Commit 91d3c2f5 authored by Erik Wilson's avatar Erik Wilson

Update runc to rc10

parent 99060a7b
...@@ -26,6 +26,7 @@ replace ( ...@@ -26,6 +26,7 @@ replace (
github.com/golangci/lint-1 => github.com/golangci/lint-1 v0.0.0-20190420132249-ee948d087217 github.com/golangci/lint-1 => github.com/golangci/lint-1 v0.0.0-20190420132249-ee948d087217
github.com/kubernetes-sigs/cri-tools => github.com/rancher/cri-tools v1.17.0-k3s1 github.com/kubernetes-sigs/cri-tools => github.com/rancher/cri-tools v1.17.0-k3s1
github.com/matryer/moq => github.com/rancher/moq v0.0.0-20190404221404-ee5226d43009 github.com/matryer/moq => github.com/rancher/moq v0.0.0-20190404221404-ee5226d43009
github.com/opencontainers/runc => github.com/opencontainers/runc v1.0.0-rc10
github.com/opencontainers/runtime-spec => github.com/opencontainers/runtime-spec v0.0.0-20180911193056-5684b8af48c1 github.com/opencontainers/runtime-spec => github.com/opencontainers/runtime-spec v0.0.0-20180911193056-5684b8af48c1
github.com/prometheus/client_golang => github.com/prometheus/client_golang v0.9.2 github.com/prometheus/client_golang => github.com/prometheus/client_golang v0.9.2
github.com/prometheus/client_model => github.com/prometheus/client_model v0.0.0-20180712105110-5c3871d89910 github.com/prometheus/client_model => github.com/prometheus/client_model v0.0.0-20180712105110-5c3871d89910
...@@ -65,6 +66,7 @@ require ( ...@@ -65,6 +66,7 @@ require (
github.com/bronze1man/goStrongswanVici v0.0.0-20190828090544-27d02f80ba40 // indirect github.com/bronze1man/goStrongswanVici v0.0.0-20190828090544-27d02f80ba40 // indirect
github.com/buger/jsonparser v0.0.0-20181115193947-bf1c66bbce23 // indirect github.com/buger/jsonparser v0.0.0-20181115193947-bf1c66bbce23 // indirect
github.com/canonical/go-dqlite v1.5.1 github.com/canonical/go-dqlite v1.5.1
github.com/cilium/ebpf v0.0.0-20200501092852-53d52285d8b3 // indirect
github.com/containerd/cgroups v0.0.0-00010101000000-000000000000 // indirect github.com/containerd/cgroups v0.0.0-00010101000000-000000000000 // indirect
github.com/containerd/containerd v1.3.0-beta.2.0.20190828155532-0293cbd26c69 github.com/containerd/containerd v1.3.0-beta.2.0.20190828155532-0293cbd26c69
github.com/containerd/continuity v0.0.0-20190827140505-75bee3e2ccb6 // indirect github.com/containerd/continuity v0.0.0-20190827140505-75bee3e2ccb6 // indirect
...@@ -110,7 +112,7 @@ require ( ...@@ -110,7 +112,7 @@ require (
github.com/urfave/cli v1.22.2 github.com/urfave/cli v1.22.2
golang.org/x/crypto v0.0.0-20200220183623-bac4c82f6975 golang.org/x/crypto v0.0.0-20200220183623-bac4c82f6975
golang.org/x/net v0.0.0-20191204025024-5ee1b9f4859a golang.org/x/net v0.0.0-20191204025024-5ee1b9f4859a
golang.org/x/sys v0.0.0-20191204072324-ce4227a45e2e golang.org/x/sys v0.0.0-20200124204421-9fbb57f87de9
google.golang.org/grpc v1.25.1 google.golang.org/grpc v1.25.1
gopkg.in/mgo.v2 v2.0.0-20190816093944-a6b53ec6cb22 // indirect gopkg.in/mgo.v2 v2.0.0-20190816093944-a6b53ec6cb22 // indirect
gopkg.in/robfig/cron.v2 v2.0.0-20150107220207-be2e0b0deed5 // indirect gopkg.in/robfig/cron.v2 v2.0.0-20150107220207-be2e0b0deed5 // indirect
......
...@@ -113,9 +113,6 @@ github.com/buger/jsonparser v0.0.0-20180808090653-f4dd9f5a6b44/go.mod h1:bbYlZJ7 ...@@ -113,9 +113,6 @@ github.com/buger/jsonparser v0.0.0-20180808090653-f4dd9f5a6b44/go.mod h1:bbYlZJ7
github.com/buger/jsonparser v0.0.0-20181115193947-bf1c66bbce23 h1:D21IyuvjDCshj1/qq+pCNd3VZOAEI9jy6Bi131YlXgI= github.com/buger/jsonparser v0.0.0-20181115193947-bf1c66bbce23 h1:D21IyuvjDCshj1/qq+pCNd3VZOAEI9jy6Bi131YlXgI=
github.com/buger/jsonparser v0.0.0-20181115193947-bf1c66bbce23/go.mod h1:bbYlZJ7hK1yFx9hf58LP0zeX7UjIGs20ufpu3evjr+s= github.com/buger/jsonparser v0.0.0-20181115193947-bf1c66bbce23/go.mod h1:bbYlZJ7hK1yFx9hf58LP0zeX7UjIGs20ufpu3evjr+s=
github.com/caddyserver/caddy v1.0.3/go.mod h1:G+ouvOY32gENkJC+jhgl62TyhvqEsFaDiZ4uw0RzP1E= github.com/caddyserver/caddy v1.0.3/go.mod h1:G+ouvOY32gENkJC+jhgl62TyhvqEsFaDiZ4uw0RzP1E=
github.com/canonical/go-dqlite v1.2.0/go.mod h1:wp00vfMvPYgNCyxcPdHB5XExmDoCGoPUGymloAQT17Y=
github.com/canonical/go-dqlite v1.3.0 h1:c+7eGZfh0K7yCmGrBkNRGZdY8R8+2jSSkz6Zr3YCjJE=
github.com/canonical/go-dqlite v1.3.0/go.mod h1:wp00vfMvPYgNCyxcPdHB5XExmDoCGoPUGymloAQT17Y=
github.com/canonical/go-dqlite v1.5.1 h1:1YjtIrFsC1A3XlgsX38ARAiKhvkZS63PqsEd8z3T4yU= github.com/canonical/go-dqlite v1.5.1 h1:1YjtIrFsC1A3XlgsX38ARAiKhvkZS63PqsEd8z3T4yU=
github.com/canonical/go-dqlite v1.5.1/go.mod h1:wp00vfMvPYgNCyxcPdHB5XExmDoCGoPUGymloAQT17Y= github.com/canonical/go-dqlite v1.5.1/go.mod h1:wp00vfMvPYgNCyxcPdHB5XExmDoCGoPUGymloAQT17Y=
github.com/cenkalti/backoff v2.1.1+incompatible/go.mod h1:90ReRw6GdpyfrHakVjL/QHaoyV4aDUVVkXQJJJ3NXXM= github.com/cenkalti/backoff v2.1.1+incompatible/go.mod h1:90ReRw6GdpyfrHakVjL/QHaoyV4aDUVVkXQJJJ3NXXM=
...@@ -127,6 +124,8 @@ github.com/chai2010/gettext-go v0.0.0-20160711120539-c6fed771bfd5/go.mod h1:/iP1 ...@@ -127,6 +124,8 @@ github.com/chai2010/gettext-go v0.0.0-20160711120539-c6fed771bfd5/go.mod h1:/iP1
github.com/checkpoint-restore/go-criu v0.0.0-20190109184317-bdb7599cd87b h1:T4nWG1TXIxeor8mAu5bFguPJgSIGhZqv/f0z55KCrJM= github.com/checkpoint-restore/go-criu v0.0.0-20190109184317-bdb7599cd87b h1:T4nWG1TXIxeor8mAu5bFguPJgSIGhZqv/f0z55KCrJM=
github.com/checkpoint-restore/go-criu v0.0.0-20190109184317-bdb7599cd87b/go.mod h1:TrMrLQfeENAPYPRsJuq3jsqdlRh3lvi6trTZJG8+tho= github.com/checkpoint-restore/go-criu v0.0.0-20190109184317-bdb7599cd87b/go.mod h1:TrMrLQfeENAPYPRsJuq3jsqdlRh3lvi6trTZJG8+tho=
github.com/cheekybits/genny v0.0.0-20170328200008-9127e812e1e9/go.mod h1:+tQajlRqAUrPI7DOSpB0XAqZYtQakVtB7wXkRAgjxjQ= github.com/cheekybits/genny v0.0.0-20170328200008-9127e812e1e9/go.mod h1:+tQajlRqAUrPI7DOSpB0XAqZYtQakVtB7wXkRAgjxjQ=
github.com/cilium/ebpf v0.0.0-20200501092852-53d52285d8b3 h1:ycJILTuzypoGfRnSbZmMuPlFVnXy6uk3CFy5QEwD2aI=
github.com/cilium/ebpf v0.0.0-20200501092852-53d52285d8b3/go.mod h1:XT+cAw5wfvsodedcijoh1l9cf7v1x9FlFB/3VmF/O8s=
github.com/client9/misspell v0.3.4/go.mod h1:qj6jICC3Q7zFZvVWo7KLAzC3yx5G7kyvSDkc90ppPyw= github.com/client9/misspell v0.3.4/go.mod h1:qj6jICC3Q7zFZvVWo7KLAzC3yx5G7kyvSDkc90ppPyw=
github.com/cloudflare/cloudflare-go v0.8.5/go.mod h1:8KhU6K+zHUEWOSU++mEQYf7D9UZOcQcibUoSm6vCUz4= github.com/cloudflare/cloudflare-go v0.8.5/go.mod h1:8KhU6K+zHUEWOSU++mEQYf7D9UZOcQcibUoSm6vCUz4=
github.com/clusterhq/flocker-go v0.0.0-20160920122132-2b8b7259d313/go.mod h1:P1wt9Z3DP8O6W3rvwCt0REIlshg1InHImaLW0t3ObY0= github.com/clusterhq/flocker-go v0.0.0-20160920122132-2b8b7259d313/go.mod h1:P1wt9Z3DP8O6W3rvwCt0REIlshg1InHImaLW0t3ObY0=
...@@ -669,10 +668,8 @@ github.com/opencontainers/go-digest v1.0.0-rc1 h1:WzifXhOVOEOuFYOJAW6aQqW0TooG2i ...@@ -669,10 +668,8 @@ github.com/opencontainers/go-digest v1.0.0-rc1 h1:WzifXhOVOEOuFYOJAW6aQqW0TooG2i
github.com/opencontainers/go-digest v1.0.0-rc1/go.mod h1:cMLVZDEM3+U2I4VmLI6N8jQYUd2OVphdqWwCJHrFt2s= github.com/opencontainers/go-digest v1.0.0-rc1/go.mod h1:cMLVZDEM3+U2I4VmLI6N8jQYUd2OVphdqWwCJHrFt2s=
github.com/opencontainers/image-spec v1.0.1 h1:JMemWkRwHx4Zj+fVxWoMCFm/8sYGGrUVojFA6h/TRcI= github.com/opencontainers/image-spec v1.0.1 h1:JMemWkRwHx4Zj+fVxWoMCFm/8sYGGrUVojFA6h/TRcI=
github.com/opencontainers/image-spec v1.0.1/go.mod h1:BtxoFyWECRxE4U/7sNtV5W15zMzWCbyJoFRP3s7yZA0= github.com/opencontainers/image-spec v1.0.1/go.mod h1:BtxoFyWECRxE4U/7sNtV5W15zMzWCbyJoFRP3s7yZA0=
github.com/opencontainers/runc v0.0.0-20190115041553-12f6a991201f/go.mod h1:qT5XzbpPznkRYVz/mWwUaVBUv2rmF59PVA73FjuZG0U= github.com/opencontainers/runc v1.0.0-rc10 h1:AbmCEuSZXVflng0/cboQkpdEOeBsPMjz6tmq4Pv8MZw=
github.com/opencontainers/runc v0.1.1/go.mod h1:qT5XzbpPznkRYVz/mWwUaVBUv2rmF59PVA73FjuZG0U= github.com/opencontainers/runc v1.0.0-rc10/go.mod h1:qT5XzbpPznkRYVz/mWwUaVBUv2rmF59PVA73FjuZG0U=
github.com/opencontainers/runc v1.0.0-rc9 h1:/k06BMULKF5hidyoZymkoDCzdJzltZpz/UU4LguQVtc=
github.com/opencontainers/runc v1.0.0-rc9/go.mod h1:qT5XzbpPznkRYVz/mWwUaVBUv2rmF59PVA73FjuZG0U=
github.com/opencontainers/runtime-spec v0.0.0-20180911193056-5684b8af48c1 h1:AcjCvgprf9I23wEYTHuyuHcuuQAp4hK5l+u1FUXgVaM= github.com/opencontainers/runtime-spec v0.0.0-20180911193056-5684b8af48c1 h1:AcjCvgprf9I23wEYTHuyuHcuuQAp4hK5l+u1FUXgVaM=
github.com/opencontainers/runtime-spec v0.0.0-20180911193056-5684b8af48c1/go.mod h1:jwyrGlmzljRJv/Fgzds9SsS/C5hL+LL3ko9hs6T5lQ0= github.com/opencontainers/runtime-spec v0.0.0-20180911193056-5684b8af48c1/go.mod h1:jwyrGlmzljRJv/Fgzds9SsS/C5hL+LL3ko9hs6T5lQ0=
github.com/opencontainers/runtime-tools v0.0.0-20181011054405-1d69bd0f9c39/go.mod h1:r3f7wjNzSs2extwzU3Y+6pKfobzPh+kKFJ3ofN+3nfs= github.com/opencontainers/runtime-tools v0.0.0-20181011054405-1d69bd0f9c39/go.mod h1:r3f7wjNzSs2extwzU3Y+6pKfobzPh+kKFJ3ofN+3nfs=
...@@ -722,10 +719,6 @@ github.com/rancher/flannel v0.11.0-k3s.1 h1:mIwnfWDafjzQgFkZeJ1AkFrrAT3EdBaA1giE ...@@ -722,10 +719,6 @@ github.com/rancher/flannel v0.11.0-k3s.1 h1:mIwnfWDafjzQgFkZeJ1AkFrrAT3EdBaA1giE
github.com/rancher/flannel v0.11.0-k3s.1/go.mod h1:Hn4ZV+eq0LhLZP63xZnxdGwXEoRSxs5sxELxu27M3UA= github.com/rancher/flannel v0.11.0-k3s.1/go.mod h1:Hn4ZV+eq0LhLZP63xZnxdGwXEoRSxs5sxELxu27M3UA=
github.com/rancher/helm-controller v0.4.1 h1:AbI3zXy2Qtw4e4A6IkxKozgnm6T3KKnUo6EzD1Pz0aU= github.com/rancher/helm-controller v0.4.1 h1:AbI3zXy2Qtw4e4A6IkxKozgnm6T3KKnUo6EzD1Pz0aU=
github.com/rancher/helm-controller v0.4.1/go.mod h1:194LHuZRrxcD82bG1rJtOWsw98U4JbPhDWqvL7l3PAw= github.com/rancher/helm-controller v0.4.1/go.mod h1:194LHuZRrxcD82bG1rJtOWsw98U4JbPhDWqvL7l3PAw=
github.com/rancher/kine v0.3.5 h1:Tm4eOtejpnzs1WFBrXj76lCLvX9czLlTkgqUk9luCQk=
github.com/rancher/kine v0.3.5/go.mod h1:xEMl0tLCva9/9me7mXJ3m9Vo6yqHgC4OU3NiK4CPrGQ=
github.com/rancher/kine v0.3.6-0.20200422224205-0a0f5b924129 h1:4HYvCG8+pfkBlpYLv/5ZOdxagg3jTszMOqMjamMQ0hA=
github.com/rancher/kine v0.3.6-0.20200422224205-0a0f5b924129/go.mod h1:xEMl0tLCva9/9me7mXJ3m9Vo6yqHgC4OU3NiK4CPrGQ=
github.com/rancher/kine v0.4.0 h1:1IhWy3TzjExG8xnj46eyUEWdzqNAD1WrgL4eEBKm6Uc= github.com/rancher/kine v0.4.0 h1:1IhWy3TzjExG8xnj46eyUEWdzqNAD1WrgL4eEBKm6Uc=
github.com/rancher/kine v0.4.0/go.mod h1:IImtCJ68AIkE+VY/kUI0NkyJL5q5WzO8QvMsSXqbrpA= github.com/rancher/kine v0.4.0/go.mod h1:IImtCJ68AIkE+VY/kUI0NkyJL5q5WzO8QvMsSXqbrpA=
github.com/rancher/kubernetes v1.17.5-k3s1 h1:rE+lSQ7SszfLZ4KB94FPQtvqQPYoU6Jwul8HHBPqQcY= github.com/rancher/kubernetes v1.17.5-k3s1 h1:rE+lSQ7SszfLZ4KB94FPQtvqQPYoU6Jwul8HHBPqQcY=
...@@ -1031,6 +1024,8 @@ golang.org/x/sys v0.0.0-20190916202348-b4ddaad3f8a3/go.mod h1:h1NjWce9XRLGQEsW7w ...@@ -1031,6 +1024,8 @@ golang.org/x/sys v0.0.0-20190916202348-b4ddaad3f8a3/go.mod h1:h1NjWce9XRLGQEsW7w
golang.org/x/sys v0.0.0-20191010194322-b09406accb47/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs= golang.org/x/sys v0.0.0-20191010194322-b09406accb47/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
golang.org/x/sys v0.0.0-20191204072324-ce4227a45e2e h1:9vRrk9YW2BTzLP0VCB9ZDjU4cPqkg+IDWL7XgxA1yxQ= golang.org/x/sys v0.0.0-20191204072324-ce4227a45e2e h1:9vRrk9YW2BTzLP0VCB9ZDjU4cPqkg+IDWL7XgxA1yxQ=
golang.org/x/sys v0.0.0-20191204072324-ce4227a45e2e/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs= golang.org/x/sys v0.0.0-20191204072324-ce4227a45e2e/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
golang.org/x/sys v0.0.0-20200124204421-9fbb57f87de9 h1:1/DFK4b7JH8DmkqhUk48onnSfrPzImPoVxuomtbT2nk=
golang.org/x/sys v0.0.0-20200124204421-9fbb57f87de9/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
golang.org/x/text v0.0.0-20160726164857-2910a502d2bf/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ= golang.org/x/text v0.0.0-20160726164857-2910a502d2bf/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
golang.org/x/text v0.0.0-20170915090833-1cbadb444a80/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ= golang.org/x/text v0.0.0-20170915090833-1cbadb444a80/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ= golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
...@@ -1078,6 +1073,8 @@ golang.org/x/tools v0.0.0-20191017205301-920acffc3e65 h1:GwXwgmbrvlcHLDsENMqrQTT ...@@ -1078,6 +1073,8 @@ golang.org/x/tools v0.0.0-20191017205301-920acffc3e65 h1:GwXwgmbrvlcHLDsENMqrQTT
golang.org/x/tools v0.0.0-20191017205301-920acffc3e65/go.mod h1:b+2E5dAYhXwXZwtnZ6UAqBI28+e2cm9otk0dWdXHAEo= golang.org/x/tools v0.0.0-20191017205301-920acffc3e65/go.mod h1:b+2E5dAYhXwXZwtnZ6UAqBI28+e2cm9otk0dWdXHAEo=
golang.org/x/xerrors v0.0.0-20190717185122-a985d3407aa7/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0= golang.org/x/xerrors v0.0.0-20190717185122-a985d3407aa7/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
golang.org/x/xerrors v0.0.0-20191011141410-1b5146add898/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0= golang.org/x/xerrors v0.0.0-20191011141410-1b5146add898/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543 h1:E7g+9GITq07hpfrRu66IVDexMakfv52eLZ2CXBWiKr4=
golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
gomodules.xyz/jsonpatch/v2 v2.0.1/go.mod h1:IhYNNY4jnS53ZnfE4PAmpKtDpTCj1JFXc+3mwe7XcUU= gomodules.xyz/jsonpatch/v2 v2.0.1/go.mod h1:IhYNNY4jnS53ZnfE4PAmpKtDpTCj1JFXc+3mwe7XcUU=
gonum.org/v1/gonum v0.0.0-20190331200053-3d26580ed485 h1:OB/uP/Puiu5vS5QMRPrXCDWUPb+kt8f1KW8oQzFejQw= gonum.org/v1/gonum v0.0.0-20190331200053-3d26580ed485 h1:OB/uP/Puiu5vS5QMRPrXCDWUPb+kt8f1KW8oQzFejQw=
gonum.org/v1/gonum v0.0.0-20190331200053-3d26580ed485/go.mod h1:2ltnJ7xHfj0zHS40VVPYEAAMTa3ZGguvHGBSJeRWqE0= gonum.org/v1/gonum v0.0.0-20190331200053-3d26580ed485/go.mod h1:2ltnJ7xHfj0zHS40VVPYEAAMTa3ZGguvHGBSJeRWqE0=
......
# Binaries for programs and plugins
*.exe
*.exe~
*.dll
*.so
*.dylib
# Test binary, build with `go test -c`
*.test
# Output of the go coverage tool, specifically when used with LiteIDE
*.out
# Contributor Covenant Code of Conduct
## Our Pledge
In the interest of fostering an open and welcoming environment, we as contributors and maintainers pledge to making participation in our project and our community a harassment-free experience for everyone, regardless of age, body size, disability, ethnicity, gender identity and expression, level of experience, nationality, personal appearance, race, religion, or sexual identity and orientation.
## Our Standards
Examples of behavior that contributes to creating a positive environment include:
* Using welcoming and inclusive language
* Being respectful of differing viewpoints and experiences
* Gracefully accepting constructive criticism
* Focusing on what is best for the community
* Showing empathy towards other community members
Examples of unacceptable behavior by participants include:
* The use of sexualized language or imagery and unwelcome sexual attention or advances
* Trolling, insulting/derogatory comments, and personal or political attacks
* Public or private harassment
* Publishing others' private information, such as a physical or electronic address, without explicit permission
* Other conduct which could reasonably be considered inappropriate in a professional setting
## Our Responsibilities
Project maintainers are responsible for clarifying the standards of acceptable behavior and are expected to take appropriate and fair corrective action in response to any instances of unacceptable behavior.
Project maintainers have the right and responsibility to remove, edit, or reject comments, commits, code, wiki edits, issues, and other contributions that are not aligned to this Code of Conduct, or to ban temporarily or permanently any contributor for other behaviors that they deem inappropriate, threatening, offensive, or harmful.
## Scope
This Code of Conduct applies both within project spaces and in public spaces when an individual is representing the project or its community. Examples of representing a project or community include using an official project e-mail address, posting via an official social media account, or acting as an appointed representative at an online or offline event. Representation of a project may be further defined and clarified by project maintainers.
## Enforcement
Instances of abusive, harassing, or otherwise unacceptable behavior may be reported by contacting the project team at nathanjsweet at gmail dot com or i at lmb dot io. The project team will review and investigate all complaints, and will respond in a way that it deems appropriate to the circumstances. The project team is obligated to maintain confidentiality with regard to the reporter of an incident. Further details of specific enforcement policies may be posted separately.
Project maintainers who do not follow or enforce the Code of Conduct in good faith may face temporary or permanent repercussions as determined by other members of the project's leadership.
## Attribution
This Code of Conduct is adapted from the [Contributor Covenant][homepage], version 1.4, available at [http://contributor-covenant.org/version/1/4][version]
[homepage]: http://contributor-covenant.org
[version]: http://contributor-covenant.org/version/1/4/
MIT License
Copyright (c) 2017 Nathan Sweet
Copyright (c) 2018, 2019 Cloudflare
Copyright (c) 2019 Authors of Cilium
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
package ebpf
import (
"bufio"
"bytes"
"fmt"
"io"
"os"
"syscall"
"github.com/cilium/ebpf/internal"
"golang.org/x/xerrors"
)
// MapABI are the attributes of a Map which are available across all supported kernels.
type MapABI struct {
Type MapType
KeySize uint32
ValueSize uint32
MaxEntries uint32
Flags uint32
}
func newMapABIFromSpec(spec *MapSpec) *MapABI {
return &MapABI{
spec.Type,
spec.KeySize,
spec.ValueSize,
spec.MaxEntries,
spec.Flags,
}
}
func newMapABIFromFd(fd *internal.FD) (string, *MapABI, error) {
info, err := bpfGetMapInfoByFD(fd)
if err != nil {
if xerrors.Is(err, syscall.EINVAL) {
abi, err := newMapABIFromProc(fd)
return "", abi, err
}
return "", nil, err
}
return "", &MapABI{
MapType(info.mapType),
info.keySize,
info.valueSize,
info.maxEntries,
info.flags,
}, nil
}
func newMapABIFromProc(fd *internal.FD) (*MapABI, error) {
var abi MapABI
err := scanFdInfo(fd, map[string]interface{}{
"map_type": &abi.Type,
"key_size": &abi.KeySize,
"value_size": &abi.ValueSize,
"max_entries": &abi.MaxEntries,
"map_flags": &abi.Flags,
})
if err != nil {
return nil, err
}
return &abi, nil
}
// Equal returns true if two ABIs have the same values.
func (abi *MapABI) Equal(other *MapABI) bool {
switch {
case abi.Type != other.Type:
return false
case abi.KeySize != other.KeySize:
return false
case abi.ValueSize != other.ValueSize:
return false
case abi.MaxEntries != other.MaxEntries:
return false
case abi.Flags != other.Flags:
return false
default:
return true
}
}
// ProgramABI are the attributes of a Program which are available across all supported kernels.
type ProgramABI struct {
Type ProgramType
}
func newProgramABIFromSpec(spec *ProgramSpec) *ProgramABI {
return &ProgramABI{
spec.Type,
}
}
func newProgramABIFromFd(fd *internal.FD) (string, *ProgramABI, error) {
info, err := bpfGetProgInfoByFD(fd)
if err != nil {
if xerrors.Is(err, syscall.EINVAL) {
return newProgramABIFromProc(fd)
}
return "", nil, err
}
var name string
if bpfName := internal.CString(info.name[:]); bpfName != "" {
name = bpfName
} else {
name = internal.CString(info.tag[:])
}
return name, &ProgramABI{
Type: ProgramType(info.progType),
}, nil
}
func newProgramABIFromProc(fd *internal.FD) (string, *ProgramABI, error) {
var (
abi ProgramABI
name string
)
err := scanFdInfo(fd, map[string]interface{}{
"prog_type": &abi.Type,
"prog_tag": &name,
})
if xerrors.Is(err, errMissingFields) {
return "", nil, &internal.UnsupportedFeatureError{
Name: "reading ABI from /proc/self/fdinfo",
MinimumVersion: internal.Version{4, 11, 0},
}
}
if err != nil {
return "", nil, err
}
return name, &abi, nil
}
func scanFdInfo(fd *internal.FD, fields map[string]interface{}) error {
raw, err := fd.Value()
if err != nil {
return err
}
fh, err := os.Open(fmt.Sprintf("/proc/self/fdinfo/%d", raw))
if err != nil {
return err
}
defer fh.Close()
if err := scanFdInfoReader(fh, fields); err != nil {
return xerrors.Errorf("%s: %w", fh.Name(), err)
}
return nil
}
var errMissingFields = xerrors.New("missing fields")
func scanFdInfoReader(r io.Reader, fields map[string]interface{}) error {
var (
scanner = bufio.NewScanner(r)
scanned int
)
for scanner.Scan() {
parts := bytes.SplitN(scanner.Bytes(), []byte("\t"), 2)
if len(parts) != 2 {
continue
}
name := bytes.TrimSuffix(parts[0], []byte(":"))
field, ok := fields[string(name)]
if !ok {
continue
}
if n, err := fmt.Fscanln(bytes.NewReader(parts[1]), field); err != nil || n != 1 {
return xerrors.Errorf("can't parse field %s: %v", name, err)
}
scanned++
}
if err := scanner.Err(); err != nil {
return err
}
if scanned != len(fields) {
return errMissingFields
}
return nil
}
// Equal returns true if two ABIs have the same values.
func (abi *ProgramABI) Equal(other *ProgramABI) bool {
switch {
case abi.Type != other.Type:
return false
default:
return true
}
}
package asm
//go:generate stringer -output alu_string.go -type=Source,Endianness,ALUOp
// Source of ALU / ALU64 / Branch operations
//
// msb lsb
// +----+-+---+
// |op |S|cls|
// +----+-+---+
type Source uint8
const sourceMask OpCode = 0x08
// Source bitmask
const (
// InvalidSource is returned by getters when invoked
// on non ALU / branch OpCodes.
InvalidSource Source = 0xff
// ImmSource src is from constant
ImmSource Source = 0x00
// RegSource src is from register
RegSource Source = 0x08
)
// The Endianness of a byte swap instruction.
type Endianness uint8
const endianMask = sourceMask
// Endian flags
const (
InvalidEndian Endianness = 0xff
// Convert to little endian
LE Endianness = 0x00
// Convert to big endian
BE Endianness = 0x08
)
// ALUOp are ALU / ALU64 operations
//
// msb lsb
// +----+-+---+
// |OP |s|cls|
// +----+-+---+
type ALUOp uint8
const aluMask OpCode = 0xf0
const (
// InvalidALUOp is returned by getters when invoked
// on non ALU OpCodes
InvalidALUOp ALUOp = 0xff
// Add - addition
Add ALUOp = 0x00
// Sub - subtraction
Sub ALUOp = 0x10
// Mul - multiplication
Mul ALUOp = 0x20
// Div - division
Div ALUOp = 0x30
// Or - bitwise or
Or ALUOp = 0x40
// And - bitwise and
And ALUOp = 0x50
// LSh - bitwise shift left
LSh ALUOp = 0x60
// RSh - bitwise shift right
RSh ALUOp = 0x70
// Neg - sign/unsign signing bit
Neg ALUOp = 0x80
// Mod - modulo
Mod ALUOp = 0x90
// Xor - bitwise xor
Xor ALUOp = 0xa0
// Mov - move value from one place to another
Mov ALUOp = 0xb0
// ArSh - arithmatic shift
ArSh ALUOp = 0xc0
// Swap - endian conversions
Swap ALUOp = 0xd0
)
// HostTo converts from host to another endianness.
func HostTo(endian Endianness, dst Register, size Size) Instruction {
var imm int64
switch size {
case Half:
imm = 16
case Word:
imm = 32
case DWord:
imm = 64
default:
return Instruction{OpCode: InvalidOpCode}
}
return Instruction{
OpCode: OpCode(ALUClass).SetALUOp(Swap).SetSource(Source(endian)),
Dst: dst,
Constant: imm,
}
}
// Op returns the OpCode for an ALU operation with a given source.
func (op ALUOp) Op(source Source) OpCode {
return OpCode(ALU64Class).SetALUOp(op).SetSource(source)
}
// Reg emits `dst (op) src`.
func (op ALUOp) Reg(dst, src Register) Instruction {
return Instruction{
OpCode: op.Op(RegSource),
Dst: dst,
Src: src,
}
}
// Imm emits `dst (op) value`.
func (op ALUOp) Imm(dst Register, value int32) Instruction {
return Instruction{
OpCode: op.Op(ImmSource),
Dst: dst,
Constant: int64(value),
}
}
// Op32 returns the OpCode for a 32-bit ALU operation with a given source.
func (op ALUOp) Op32(source Source) OpCode {
return OpCode(ALUClass).SetALUOp(op).SetSource(source)
}
// Reg32 emits `dst (op) src`, zeroing the upper 32 bit of dst.
func (op ALUOp) Reg32(dst, src Register) Instruction {
return Instruction{
OpCode: op.Op32(RegSource),
Dst: dst,
Src: src,
}
}
// Imm32 emits `dst (op) value`, zeroing the upper 32 bit of dst.
func (op ALUOp) Imm32(dst Register, value int32) Instruction {
return Instruction{
OpCode: op.Op32(ImmSource),
Dst: dst,
Constant: int64(value),
}
}
// Code generated by "stringer -output alu_string.go -type=Source,Endianness,ALUOp"; DO NOT EDIT.
package asm
import "strconv"
func _() {
// An "invalid array index" compiler error signifies that the constant values have changed.
// Re-run the stringer command to generate them again.
var x [1]struct{}
_ = x[InvalidSource-255]
_ = x[ImmSource-0]
_ = x[RegSource-8]
}
const (
_Source_name_0 = "ImmSource"
_Source_name_1 = "RegSource"
_Source_name_2 = "InvalidSource"
)
func (i Source) String() string {
switch {
case i == 0:
return _Source_name_0
case i == 8:
return _Source_name_1
case i == 255:
return _Source_name_2
default:
return "Source(" + strconv.FormatInt(int64(i), 10) + ")"
}
}
func _() {
// An "invalid array index" compiler error signifies that the constant values have changed.
// Re-run the stringer command to generate them again.
var x [1]struct{}
_ = x[InvalidEndian-255]
_ = x[LE-0]
_ = x[BE-8]
}
const (
_Endianness_name_0 = "LE"
_Endianness_name_1 = "BE"
_Endianness_name_2 = "InvalidEndian"
)
func (i Endianness) String() string {
switch {
case i == 0:
return _Endianness_name_0
case i == 8:
return _Endianness_name_1
case i == 255:
return _Endianness_name_2
default:
return "Endianness(" + strconv.FormatInt(int64(i), 10) + ")"
}
}
func _() {
// An "invalid array index" compiler error signifies that the constant values have changed.
// Re-run the stringer command to generate them again.
var x [1]struct{}
_ = x[InvalidALUOp-255]
_ = x[Add-0]
_ = x[Sub-16]
_ = x[Mul-32]
_ = x[Div-48]
_ = x[Or-64]
_ = x[And-80]
_ = x[LSh-96]
_ = x[RSh-112]
_ = x[Neg-128]
_ = x[Mod-144]
_ = x[Xor-160]
_ = x[Mov-176]
_ = x[ArSh-192]
_ = x[Swap-208]
}
const _ALUOp_name = "AddSubMulDivOrAndLShRShNegModXorMovArShSwapInvalidALUOp"
var _ALUOp_map = map[ALUOp]string{
0: _ALUOp_name[0:3],
16: _ALUOp_name[3:6],
32: _ALUOp_name[6:9],
48: _ALUOp_name[9:12],
64: _ALUOp_name[12:14],
80: _ALUOp_name[14:17],
96: _ALUOp_name[17:20],
112: _ALUOp_name[20:23],
128: _ALUOp_name[23:26],
144: _ALUOp_name[26:29],
160: _ALUOp_name[29:32],
176: _ALUOp_name[32:35],
192: _ALUOp_name[35:39],
208: _ALUOp_name[39:43],
255: _ALUOp_name[43:55],
}
func (i ALUOp) String() string {
if str, ok := _ALUOp_map[i]; ok {
return str
}
return "ALUOp(" + strconv.FormatInt(int64(i), 10) + ")"
}
// Package asm is an assembler for eBPF bytecode.
package asm
package asm
//go:generate stringer -output func_string.go -type=BuiltinFunc
// BuiltinFunc is a built-in eBPF function.
type BuiltinFunc int32
// eBPF built-in functions
//
// You can renegerate this list using the following gawk script:
//
// /FN\(.+\),/ {
// match($1, /\((.+)\)/, r)
// split(r[1], p, "_")
// printf "Fn"
// for (i in p) {
// printf "%s%s", toupper(substr(p[i], 1, 1)), substr(p[i], 2)
// }
// print ""
// }
//
// The script expects include/uapi/linux/bpf.h as it's input.
const (
FnUnspec BuiltinFunc = iota
FnMapLookupElem
FnMapUpdateElem
FnMapDeleteElem
FnProbeRead
FnKtimeGetNs
FnTracePrintk
FnGetPrandomU32
FnGetSmpProcessorId
FnSkbStoreBytes
FnL3CsumReplace
FnL4CsumReplace
FnTailCall
FnCloneRedirect
FnGetCurrentPidTgid
FnGetCurrentUidGid
FnGetCurrentComm
FnGetCgroupClassid
FnSkbVlanPush
FnSkbVlanPop
FnSkbGetTunnelKey
FnSkbSetTunnelKey
FnPerfEventRead
FnRedirect
FnGetRouteRealm
FnPerfEventOutput
FnSkbLoadBytes
FnGetStackid
FnCsumDiff
FnSkbGetTunnelOpt
FnSkbSetTunnelOpt
FnSkbChangeProto
FnSkbChangeType
FnSkbUnderCgroup
FnGetHashRecalc
FnGetCurrentTask
FnProbeWriteUser
FnCurrentTaskUnderCgroup
FnSkbChangeTail
FnSkbPullData
FnCsumUpdate
FnSetHashInvalid
FnGetNumaNodeId
FnSkbChangeHead
FnXdpAdjustHead
FnProbeReadStr
FnGetSocketCookie
FnGetSocketUid
FnSetHash
FnSetsockopt
FnSkbAdjustRoom
FnRedirectMap
FnSkRedirectMap
FnSockMapUpdate
FnXdpAdjustMeta
FnPerfEventReadValue
FnPerfProgReadValue
FnGetsockopt
FnOverrideReturn
FnSockOpsCbFlagsSet
FnMsgRedirectMap
FnMsgApplyBytes
FnMsgCorkBytes
FnMsgPullData
FnBind
FnXdpAdjustTail
FnSkbGetXfrmState
FnGetStack
FnSkbLoadBytesRelative
FnFibLookup
FnSockHashUpdate
FnMsgRedirectHash
FnSkRedirectHash
FnLwtPushEncap
FnLwtSeg6StoreBytes
FnLwtSeg6AdjustSrh
FnLwtSeg6Action
FnRcRepeat
FnRcKeydown
FnSkbCgroupId
FnGetCurrentCgroupId
FnGetLocalStorage
FnSkSelectReuseport
FnSkbAncestorCgroupId
FnSkLookupTcp
FnSkLookupUdp
FnSkRelease
FnMapPushElem
FnMapPopElem
FnMapPeekElem
FnMsgPushData
FnMsgPopData
FnRcPointerRel
FnSpinLock
FnSpinUnlock
FnSkFullsock
FnTcpSock
FnSkbEcnSetCe
FnGetListenerSock
FnSkcLookupTcp
FnTcpCheckSyncookie
FnSysctlGetName
FnSysctlGetCurrentValue
FnSysctlGetNewValue
FnSysctlSetNewValue
FnStrtol
FnStrtoul
FnSkStorageGet
FnSkStorageDelete
FnSendSignal
FnTcpGenSyncookie
)
// Call emits a function call.
func (fn BuiltinFunc) Call() Instruction {
return Instruction{
OpCode: OpCode(JumpClass).SetJumpOp(Call),
Constant: int64(fn),
}
}
// Code generated by "stringer -output func_string.go -type=BuiltinFunc"; DO NOT EDIT.
package asm
import "strconv"
func _() {
// An "invalid array index" compiler error signifies that the constant values have changed.
// Re-run the stringer command to generate them again.
var x [1]struct{}
_ = x[FnUnspec-0]
_ = x[FnMapLookupElem-1]
_ = x[FnMapUpdateElem-2]
_ = x[FnMapDeleteElem-3]
_ = x[FnProbeRead-4]
_ = x[FnKtimeGetNs-5]
_ = x[FnTracePrintk-6]
_ = x[FnGetPrandomU32-7]
_ = x[FnGetSmpProcessorId-8]
_ = x[FnSkbStoreBytes-9]
_ = x[FnL3CsumReplace-10]
_ = x[FnL4CsumReplace-11]
_ = x[FnTailCall-12]
_ = x[FnCloneRedirect-13]
_ = x[FnGetCurrentPidTgid-14]
_ = x[FnGetCurrentUidGid-15]
_ = x[FnGetCurrentComm-16]
_ = x[FnGetCgroupClassid-17]
_ = x[FnSkbVlanPush-18]
_ = x[FnSkbVlanPop-19]
_ = x[FnSkbGetTunnelKey-20]
_ = x[FnSkbSetTunnelKey-21]
_ = x[FnPerfEventRead-22]
_ = x[FnRedirect-23]
_ = x[FnGetRouteRealm-24]
_ = x[FnPerfEventOutput-25]
_ = x[FnSkbLoadBytes-26]
_ = x[FnGetStackid-27]
_ = x[FnCsumDiff-28]
_ = x[FnSkbGetTunnelOpt-29]
_ = x[FnSkbSetTunnelOpt-30]
_ = x[FnSkbChangeProto-31]
_ = x[FnSkbChangeType-32]
_ = x[FnSkbUnderCgroup-33]
_ = x[FnGetHashRecalc-34]
_ = x[FnGetCurrentTask-35]
_ = x[FnProbeWriteUser-36]
_ = x[FnCurrentTaskUnderCgroup-37]
_ = x[FnSkbChangeTail-38]
_ = x[FnSkbPullData-39]
_ = x[FnCsumUpdate-40]
_ = x[FnSetHashInvalid-41]
_ = x[FnGetNumaNodeId-42]
_ = x[FnSkbChangeHead-43]
_ = x[FnXdpAdjustHead-44]
_ = x[FnProbeReadStr-45]
_ = x[FnGetSocketCookie-46]
_ = x[FnGetSocketUid-47]
_ = x[FnSetHash-48]
_ = x[FnSetsockopt-49]
_ = x[FnSkbAdjustRoom-50]
_ = x[FnRedirectMap-51]
_ = x[FnSkRedirectMap-52]
_ = x[FnSockMapUpdate-53]
_ = x[FnXdpAdjustMeta-54]
_ = x[FnPerfEventReadValue-55]
_ = x[FnPerfProgReadValue-56]
_ = x[FnGetsockopt-57]
_ = x[FnOverrideReturn-58]
_ = x[FnSockOpsCbFlagsSet-59]
_ = x[FnMsgRedirectMap-60]
_ = x[FnMsgApplyBytes-61]
_ = x[FnMsgCorkBytes-62]
_ = x[FnMsgPullData-63]
_ = x[FnBind-64]
_ = x[FnXdpAdjustTail-65]
_ = x[FnSkbGetXfrmState-66]
_ = x[FnGetStack-67]
_ = x[FnSkbLoadBytesRelative-68]
_ = x[FnFibLookup-69]
_ = x[FnSockHashUpdate-70]
_ = x[FnMsgRedirectHash-71]
_ = x[FnSkRedirectHash-72]
_ = x[FnLwtPushEncap-73]
_ = x[FnLwtSeg6StoreBytes-74]
_ = x[FnLwtSeg6AdjustSrh-75]
_ = x[FnLwtSeg6Action-76]
_ = x[FnRcRepeat-77]
_ = x[FnRcKeydown-78]
_ = x[FnSkbCgroupId-79]
_ = x[FnGetCurrentCgroupId-80]
_ = x[FnGetLocalStorage-81]
_ = x[FnSkSelectReuseport-82]
_ = x[FnSkbAncestorCgroupId-83]
_ = x[FnSkLookupTcp-84]
_ = x[FnSkLookupUdp-85]
_ = x[FnSkRelease-86]
_ = x[FnMapPushElem-87]
_ = x[FnMapPopElem-88]
_ = x[FnMapPeekElem-89]
_ = x[FnMsgPushData-90]
_ = x[FnMsgPopData-91]
_ = x[FnRcPointerRel-92]
_ = x[FnSpinLock-93]
_ = x[FnSpinUnlock-94]
_ = x[FnSkFullsock-95]
_ = x[FnTcpSock-96]
_ = x[FnSkbEcnSetCe-97]
_ = x[FnGetListenerSock-98]
_ = x[FnSkcLookupTcp-99]
_ = x[FnTcpCheckSyncookie-100]
_ = x[FnSysctlGetName-101]
_ = x[FnSysctlGetCurrentValue-102]
_ = x[FnSysctlGetNewValue-103]
_ = x[FnSysctlSetNewValue-104]
_ = x[FnStrtol-105]
_ = x[FnStrtoul-106]
_ = x[FnSkStorageGet-107]
_ = x[FnSkStorageDelete-108]
_ = x[FnSendSignal-109]
_ = x[FnTcpGenSyncookie-110]
}
const _BuiltinFunc_name = "FnUnspecFnMapLookupElemFnMapUpdateElemFnMapDeleteElemFnProbeReadFnKtimeGetNsFnTracePrintkFnGetPrandomU32FnGetSmpProcessorIdFnSkbStoreBytesFnL3CsumReplaceFnL4CsumReplaceFnTailCallFnCloneRedirectFnGetCurrentPidTgidFnGetCurrentUidGidFnGetCurrentCommFnGetCgroupClassidFnSkbVlanPushFnSkbVlanPopFnSkbGetTunnelKeyFnSkbSetTunnelKeyFnPerfEventReadFnRedirectFnGetRouteRealmFnPerfEventOutputFnSkbLoadBytesFnGetStackidFnCsumDiffFnSkbGetTunnelOptFnSkbSetTunnelOptFnSkbChangeProtoFnSkbChangeTypeFnSkbUnderCgroupFnGetHashRecalcFnGetCurrentTaskFnProbeWriteUserFnCurrentTaskUnderCgroupFnSkbChangeTailFnSkbPullDataFnCsumUpdateFnSetHashInvalidFnGetNumaNodeIdFnSkbChangeHeadFnXdpAdjustHeadFnProbeReadStrFnGetSocketCookieFnGetSocketUidFnSetHashFnSetsockoptFnSkbAdjustRoomFnRedirectMapFnSkRedirectMapFnSockMapUpdateFnXdpAdjustMetaFnPerfEventReadValueFnPerfProgReadValueFnGetsockoptFnOverrideReturnFnSockOpsCbFlagsSetFnMsgRedirectMapFnMsgApplyBytesFnMsgCorkBytesFnMsgPullDataFnBindFnXdpAdjustTailFnSkbGetXfrmStateFnGetStackFnSkbLoadBytesRelativeFnFibLookupFnSockHashUpdateFnMsgRedirectHashFnSkRedirectHashFnLwtPushEncapFnLwtSeg6StoreBytesFnLwtSeg6AdjustSrhFnLwtSeg6ActionFnRcRepeatFnRcKeydownFnSkbCgroupIdFnGetCurrentCgroupIdFnGetLocalStorageFnSkSelectReuseportFnSkbAncestorCgroupIdFnSkLookupTcpFnSkLookupUdpFnSkReleaseFnMapPushElemFnMapPopElemFnMapPeekElemFnMsgPushDataFnMsgPopDataFnRcPointerRelFnSpinLockFnSpinUnlockFnSkFullsockFnTcpSockFnSkbEcnSetCeFnGetListenerSockFnSkcLookupTcpFnTcpCheckSyncookieFnSysctlGetNameFnSysctlGetCurrentValueFnSysctlGetNewValueFnSysctlSetNewValueFnStrtolFnStrtoulFnSkStorageGetFnSkStorageDeleteFnSendSignalFnTcpGenSyncookie"
var _BuiltinFunc_index = [...]uint16{0, 8, 23, 38, 53, 64, 76, 89, 104, 123, 138, 153, 168, 178, 193, 212, 230, 246, 264, 277, 289, 306, 323, 338, 348, 363, 380, 394, 406, 416, 433, 450, 466, 481, 497, 512, 528, 544, 568, 583, 596, 608, 624, 639, 654, 669, 683, 700, 714, 723, 735, 750, 763, 778, 793, 808, 828, 847, 859, 875, 894, 910, 925, 939, 952, 958, 973, 990, 1000, 1022, 1033, 1049, 1066, 1082, 1096, 1115, 1133, 1148, 1158, 1169, 1182, 1202, 1219, 1238, 1259, 1272, 1285, 1296, 1309, 1321, 1334, 1347, 1359, 1373, 1383, 1395, 1407, 1416, 1429, 1446, 1460, 1479, 1494, 1517, 1536, 1555, 1563, 1572, 1586, 1603, 1615, 1632}
func (i BuiltinFunc) String() string {
if i < 0 || i >= BuiltinFunc(len(_BuiltinFunc_index)-1) {
return "BuiltinFunc(" + strconv.FormatInt(int64(i), 10) + ")"
}
return _BuiltinFunc_name[_BuiltinFunc_index[i]:_BuiltinFunc_index[i+1]]
}
package asm
//go:generate stringer -output jump_string.go -type=JumpOp
// JumpOp affect control flow.
//
// msb lsb
// +----+-+---+
// |OP |s|cls|
// +----+-+---+
type JumpOp uint8
const jumpMask OpCode = aluMask
const (
// InvalidJumpOp is returned by getters when invoked
// on non branch OpCodes
InvalidJumpOp JumpOp = 0xff
// Ja jumps by offset unconditionally
Ja JumpOp = 0x00
// JEq jumps by offset if r == imm
JEq JumpOp = 0x10
// JGT jumps by offset if r > imm
JGT JumpOp = 0x20
// JGE jumps by offset if r >= imm
JGE JumpOp = 0x30
// JSet jumps by offset if r & imm
JSet JumpOp = 0x40
// JNE jumps by offset if r != imm
JNE JumpOp = 0x50
// JSGT jumps by offset if signed r > signed imm
JSGT JumpOp = 0x60
// JSGE jumps by offset if signed r >= signed imm
JSGE JumpOp = 0x70
// Call builtin or user defined function from imm
Call JumpOp = 0x80
// Exit ends execution, with value in r0
Exit JumpOp = 0x90
// JLT jumps by offset if r < imm
JLT JumpOp = 0xa0
// JLE jumps by offset if r <= imm
JLE JumpOp = 0xb0
// JSLT jumps by offset if signed r < signed imm
JSLT JumpOp = 0xc0
// JSLE jumps by offset if signed r <= signed imm
JSLE JumpOp = 0xd0
)
// Return emits an exit instruction.
//
// Requires a return value in R0.
func Return() Instruction {
return Instruction{
OpCode: OpCode(JumpClass).SetJumpOp(Exit),
}
}
// Op returns the OpCode for a given jump source.
func (op JumpOp) Op(source Source) OpCode {
return OpCode(JumpClass).SetJumpOp(op).SetSource(source)
}
// Imm compares dst to value, and adjusts PC by offset if the condition is fulfilled.
func (op JumpOp) Imm(dst Register, value int32, label string) Instruction {
if op == Exit || op == Call || op == Ja {
return Instruction{OpCode: InvalidOpCode}
}
return Instruction{
OpCode: OpCode(JumpClass).SetJumpOp(op).SetSource(ImmSource),
Dst: dst,
Offset: -1,
Constant: int64(value),
Reference: label,
}
}
// Reg compares dst to src, and adjusts PC by offset if the condition is fulfilled.
func (op JumpOp) Reg(dst, src Register, label string) Instruction {
if op == Exit || op == Call || op == Ja {
return Instruction{OpCode: InvalidOpCode}
}
return Instruction{
OpCode: OpCode(JumpClass).SetJumpOp(op).SetSource(RegSource),
Dst: dst,
Src: src,
Offset: -1,
Reference: label,
}
}
// Label adjusts PC to the address of the label.
func (op JumpOp) Label(label string) Instruction {
if op == Call {
return Instruction{
OpCode: OpCode(JumpClass).SetJumpOp(Call),
Src: PseudoCall,
Constant: -1,
Reference: label,
}
}
return Instruction{
OpCode: OpCode(JumpClass).SetJumpOp(op),
Offset: -1,
Reference: label,
}
}
// Code generated by "stringer -output jump_string.go -type=JumpOp"; DO NOT EDIT.
package asm
import "strconv"
func _() {
// An "invalid array index" compiler error signifies that the constant values have changed.
// Re-run the stringer command to generate them again.
var x [1]struct{}
_ = x[InvalidJumpOp-255]
_ = x[Ja-0]
_ = x[JEq-16]
_ = x[JGT-32]
_ = x[JGE-48]
_ = x[JSet-64]
_ = x[JNE-80]
_ = x[JSGT-96]
_ = x[JSGE-112]
_ = x[Call-128]
_ = x[Exit-144]
_ = x[JLT-160]
_ = x[JLE-176]
_ = x[JSLT-192]
_ = x[JSLE-208]
}
const _JumpOp_name = "JaJEqJGTJGEJSetJNEJSGTJSGECallExitJLTJLEJSLTJSLEInvalidJumpOp"
var _JumpOp_map = map[JumpOp]string{
0: _JumpOp_name[0:2],
16: _JumpOp_name[2:5],
32: _JumpOp_name[5:8],
48: _JumpOp_name[8:11],
64: _JumpOp_name[11:15],
80: _JumpOp_name[15:18],
96: _JumpOp_name[18:22],
112: _JumpOp_name[22:26],
128: _JumpOp_name[26:30],
144: _JumpOp_name[30:34],
160: _JumpOp_name[34:37],
176: _JumpOp_name[37:40],
192: _JumpOp_name[40:44],
208: _JumpOp_name[44:48],
255: _JumpOp_name[48:61],
}
func (i JumpOp) String() string {
if str, ok := _JumpOp_map[i]; ok {
return str
}
return "JumpOp(" + strconv.FormatInt(int64(i), 10) + ")"
}
package asm
//go:generate stringer -output load_store_string.go -type=Mode,Size
// Mode for load and store operations
//
// msb lsb
// +---+--+---+
// |MDE|sz|cls|
// +---+--+---+
type Mode uint8
const modeMask OpCode = 0xe0
const (
// InvalidMode is returned by getters when invoked
// on non load / store OpCodes
InvalidMode Mode = 0xff
// ImmMode - immediate value
ImmMode Mode = 0x00
// AbsMode - immediate value + offset
AbsMode Mode = 0x20
// IndMode - indirect (imm+src)
IndMode Mode = 0x40
// MemMode - load from memory
MemMode Mode = 0x60
// XAddMode - add atomically across processors.
XAddMode Mode = 0xc0
)
// Size of load and store operations
//
// msb lsb
// +---+--+---+
// |mde|SZ|cls|
// +---+--+---+
type Size uint8
const sizeMask OpCode = 0x18
const (
// InvalidSize is returned by getters when invoked
// on non load / store OpCodes
InvalidSize Size = 0xff
// DWord - double word; 64 bits
DWord Size = 0x18
// Word - word; 32 bits
Word Size = 0x00
// Half - half-word; 16 bits
Half Size = 0x08
// Byte - byte; 8 bits
Byte Size = 0x10
)
// Sizeof returns the size in bytes.
func (s Size) Sizeof() int {
switch s {
case DWord:
return 8
case Word:
return 4
case Half:
return 2
case Byte:
return 1
default:
return -1
}
}
// LoadMemOp returns the OpCode to load a value of given size from memory.
func LoadMemOp(size Size) OpCode {
return OpCode(LdXClass).SetMode(MemMode).SetSize(size)
}
// LoadMem emits `dst = *(size *)(src + offset)`.
func LoadMem(dst, src Register, offset int16, size Size) Instruction {
return Instruction{
OpCode: LoadMemOp(size),
Dst: dst,
Src: src,
Offset: offset,
}
}
// LoadImmOp returns the OpCode to load an immediate of given size.
//
// As of kernel 4.20, only DWord size is accepted.
func LoadImmOp(size Size) OpCode {
return OpCode(LdClass).SetMode(ImmMode).SetSize(size)
}
// LoadImm emits `dst = (size)value`.
//
// As of kernel 4.20, only DWord size is accepted.
func LoadImm(dst Register, value int64, size Size) Instruction {
return Instruction{
OpCode: LoadImmOp(size),
Dst: dst,
Constant: value,
}
}
// LoadMapPtr stores a pointer to a map in dst.
func LoadMapPtr(dst Register, fd int) Instruction {
if fd < 0 {
return Instruction{OpCode: InvalidOpCode}
}
return Instruction{
OpCode: LoadImmOp(DWord),
Dst: dst,
Src: PseudoMapFD,
Constant: int64(fd),
}
}
// LoadMapValue stores a pointer to the value at a certain offset of a map.
func LoadMapValue(dst Register, fd int, offset uint32) Instruction {
if fd < 0 {
return Instruction{OpCode: InvalidOpCode}
}
fdAndOffset := (uint64(offset) << 32) | uint64(uint32(fd))
return Instruction{
OpCode: LoadImmOp(DWord),
Dst: dst,
Src: PseudoMapValue,
Constant: int64(fdAndOffset),
}
}
// LoadIndOp returns the OpCode for loading a value of given size from an sk_buff.
func LoadIndOp(size Size) OpCode {
return OpCode(LdClass).SetMode(IndMode).SetSize(size)
}
// LoadInd emits `dst = ntoh(*(size *)(((sk_buff *)R6)->data + src + offset))`.
func LoadInd(dst, src Register, offset int32, size Size) Instruction {
return Instruction{
OpCode: LoadIndOp(size),
Dst: dst,
Src: src,
Constant: int64(offset),
}
}
// LoadAbsOp returns the OpCode for loading a value of given size from an sk_buff.
func LoadAbsOp(size Size) OpCode {
return OpCode(LdClass).SetMode(AbsMode).SetSize(size)
}
// LoadAbs emits `r0 = ntoh(*(size *)(((sk_buff *)R6)->data + offset))`.
func LoadAbs(offset int32, size Size) Instruction {
return Instruction{
OpCode: LoadAbsOp(size),
Dst: R0,
Constant: int64(offset),
}
}
// StoreMemOp returns the OpCode for storing a register of given size in memory.
func StoreMemOp(size Size) OpCode {
return OpCode(StXClass).SetMode(MemMode).SetSize(size)
}
// StoreMem emits `*(size *)(dst + offset) = src`
func StoreMem(dst Register, offset int16, src Register, size Size) Instruction {
return Instruction{
OpCode: StoreMemOp(size),
Dst: dst,
Src: src,
Offset: offset,
}
}
// StoreImmOp returns the OpCode for storing an immediate of given size in memory.
func StoreImmOp(size Size) OpCode {
return OpCode(StClass).SetMode(MemMode).SetSize(size)
}
// StoreImm emits `*(size *)(dst + offset) = value`.
func StoreImm(dst Register, offset int16, value int64, size Size) Instruction {
return Instruction{
OpCode: StoreImmOp(size),
Dst: dst,
Offset: offset,
Constant: value,
}
}
// StoreXAddOp returns the OpCode to atomically add a register to a value in memory.
func StoreXAddOp(size Size) OpCode {
return OpCode(StXClass).SetMode(XAddMode).SetSize(size)
}
// StoreXAdd atomically adds src to *dst.
func StoreXAdd(dst, src Register, size Size) Instruction {
return Instruction{
OpCode: StoreXAddOp(size),
Dst: dst,
Src: src,
}
}
// Code generated by "stringer -output load_store_string.go -type=Mode,Size"; DO NOT EDIT.
package asm
import "strconv"
func _() {
// An "invalid array index" compiler error signifies that the constant values have changed.
// Re-run the stringer command to generate them again.
var x [1]struct{}
_ = x[InvalidMode-255]
_ = x[ImmMode-0]
_ = x[AbsMode-32]
_ = x[IndMode-64]
_ = x[MemMode-96]
_ = x[XAddMode-192]
}
const (
_Mode_name_0 = "ImmMode"
_Mode_name_1 = "AbsMode"
_Mode_name_2 = "IndMode"
_Mode_name_3 = "MemMode"
_Mode_name_4 = "XAddMode"
_Mode_name_5 = "InvalidMode"
)
func (i Mode) String() string {
switch {
case i == 0:
return _Mode_name_0
case i == 32:
return _Mode_name_1
case i == 64:
return _Mode_name_2
case i == 96:
return _Mode_name_3
case i == 192:
return _Mode_name_4
case i == 255:
return _Mode_name_5
default:
return "Mode(" + strconv.FormatInt(int64(i), 10) + ")"
}
}
func _() {
// An "invalid array index" compiler error signifies that the constant values have changed.
// Re-run the stringer command to generate them again.
var x [1]struct{}
_ = x[InvalidSize-255]
_ = x[DWord-24]
_ = x[Word-0]
_ = x[Half-8]
_ = x[Byte-16]
}
const (
_Size_name_0 = "Word"
_Size_name_1 = "Half"
_Size_name_2 = "Byte"
_Size_name_3 = "DWord"
_Size_name_4 = "InvalidSize"
)
func (i Size) String() string {
switch {
case i == 0:
return _Size_name_0
case i == 8:
return _Size_name_1
case i == 16:
return _Size_name_2
case i == 24:
return _Size_name_3
case i == 255:
return _Size_name_4
default:
return "Size(" + strconv.FormatInt(int64(i), 10) + ")"
}
}
package asm
import (
"fmt"
"strings"
)
//go:generate stringer -output opcode_string.go -type=Class
type encoding int
const (
unknownEncoding encoding = iota
loadOrStore
jumpOrALU
)
// Class of operations
//
// msb lsb
// +---+--+---+
// | ?? |CLS|
// +---+--+---+
type Class uint8
const classMask OpCode = 0x07
const (
// LdClass load memory
LdClass Class = 0x00
// LdXClass load memory from constant
LdXClass Class = 0x01
// StClass load register from memory
StClass Class = 0x02
// StXClass load register from constant
StXClass Class = 0x03
// ALUClass arithmetic operators
ALUClass Class = 0x04
// JumpClass jump operators
JumpClass Class = 0x05
// ALU64Class arithmetic in 64 bit mode
ALU64Class Class = 0x07
)
func (cls Class) encoding() encoding {
switch cls {
case LdClass, LdXClass, StClass, StXClass:
return loadOrStore
case ALU64Class, ALUClass, JumpClass:
return jumpOrALU
default:
return unknownEncoding
}
}
// OpCode is a packed eBPF opcode.
//
// Its encoding is defined by a Class value:
//
// msb lsb
// +----+-+---+
// | ???? |CLS|
// +----+-+---+
type OpCode uint8
// InvalidOpCode is returned by setters on OpCode
const InvalidOpCode OpCode = 0xff
// marshalledInstructions returns the number of BPF instructions required
// to encode this opcode.
func (op OpCode) marshalledInstructions() int {
if op == LoadImmOp(DWord) {
return 2
}
return 1
}
func (op OpCode) isDWordLoad() bool {
return op == LoadImmOp(DWord)
}
// Class returns the class of operation.
func (op OpCode) Class() Class {
return Class(op & classMask)
}
// Mode returns the mode for load and store operations.
func (op OpCode) Mode() Mode {
if op.Class().encoding() != loadOrStore {
return InvalidMode
}
return Mode(op & modeMask)
}
// Size returns the size for load and store operations.
func (op OpCode) Size() Size {
if op.Class().encoding() != loadOrStore {
return InvalidSize
}
return Size(op & sizeMask)
}
// Source returns the source for branch and ALU operations.
func (op OpCode) Source() Source {
if op.Class().encoding() != jumpOrALU || op.ALUOp() == Swap {
return InvalidSource
}
return Source(op & sourceMask)
}
// ALUOp returns the ALUOp.
func (op OpCode) ALUOp() ALUOp {
if op.Class().encoding() != jumpOrALU {
return InvalidALUOp
}
return ALUOp(op & aluMask)
}
// Endianness returns the Endianness for a byte swap instruction.
func (op OpCode) Endianness() Endianness {
if op.ALUOp() != Swap {
return InvalidEndian
}
return Endianness(op & endianMask)
}
// JumpOp returns the JumpOp.
func (op OpCode) JumpOp() JumpOp {
if op.Class().encoding() != jumpOrALU {
return InvalidJumpOp
}
return JumpOp(op & jumpMask)
}
// SetMode sets the mode on load and store operations.
//
// Returns InvalidOpCode if op is of the wrong class.
func (op OpCode) SetMode(mode Mode) OpCode {
if op.Class().encoding() != loadOrStore || !valid(OpCode(mode), modeMask) {
return InvalidOpCode
}
return (op & ^modeMask) | OpCode(mode)
}
// SetSize sets the size on load and store operations.
//
// Returns InvalidOpCode if op is of the wrong class.
func (op OpCode) SetSize(size Size) OpCode {
if op.Class().encoding() != loadOrStore || !valid(OpCode(size), sizeMask) {
return InvalidOpCode
}
return (op & ^sizeMask) | OpCode(size)
}
// SetSource sets the source on jump and ALU operations.
//
// Returns InvalidOpCode if op is of the wrong class.
func (op OpCode) SetSource(source Source) OpCode {
if op.Class().encoding() != jumpOrALU || !valid(OpCode(source), sourceMask) {
return InvalidOpCode
}
return (op & ^sourceMask) | OpCode(source)
}
// SetALUOp sets the ALUOp on ALU operations.
//
// Returns InvalidOpCode if op is of the wrong class.
func (op OpCode) SetALUOp(alu ALUOp) OpCode {
class := op.Class()
if (class != ALUClass && class != ALU64Class) || !valid(OpCode(alu), aluMask) {
return InvalidOpCode
}
return (op & ^aluMask) | OpCode(alu)
}
// SetJumpOp sets the JumpOp on jump operations.
//
// Returns InvalidOpCode if op is of the wrong class.
func (op OpCode) SetJumpOp(jump JumpOp) OpCode {
if op.Class() != JumpClass || !valid(OpCode(jump), jumpMask) {
return InvalidOpCode
}
return (op & ^jumpMask) | OpCode(jump)
}
func (op OpCode) String() string {
var f strings.Builder
switch class := op.Class(); class {
case LdClass, LdXClass, StClass, StXClass:
f.WriteString(strings.TrimSuffix(class.String(), "Class"))
mode := op.Mode()
f.WriteString(strings.TrimSuffix(mode.String(), "Mode"))
switch op.Size() {
case DWord:
f.WriteString("DW")
case Word:
f.WriteString("W")
case Half:
f.WriteString("H")
case Byte:
f.WriteString("B")
}
case ALU64Class, ALUClass:
f.WriteString(op.ALUOp().String())
if op.ALUOp() == Swap {
// Width for Endian is controlled by Constant
f.WriteString(op.Endianness().String())
} else {
if class == ALUClass {
f.WriteString("32")
}
f.WriteString(strings.TrimSuffix(op.Source().String(), "Source"))
}
case JumpClass:
f.WriteString(op.JumpOp().String())
if jop := op.JumpOp(); jop != Exit && jop != Call {
f.WriteString(strings.TrimSuffix(op.Source().String(), "Source"))
}
default:
fmt.Fprintf(&f, "%#x", op)
}
return f.String()
}
// valid returns true if all bits in value are covered by mask.
func valid(value, mask OpCode) bool {
return value & ^mask == 0
}
// Code generated by "stringer -output opcode_string.go -type=Class"; DO NOT EDIT.
package asm
import "strconv"
func _() {
// An "invalid array index" compiler error signifies that the constant values have changed.
// Re-run the stringer command to generate them again.
var x [1]struct{}
_ = x[LdClass-0]
_ = x[LdXClass-1]
_ = x[StClass-2]
_ = x[StXClass-3]
_ = x[ALUClass-4]
_ = x[JumpClass-5]
_ = x[ALU64Class-7]
}
const (
_Class_name_0 = "LdClassLdXClassStClassStXClassALUClassJumpClass"
_Class_name_1 = "ALU64Class"
)
var (
_Class_index_0 = [...]uint8{0, 7, 15, 22, 30, 38, 47}
)
func (i Class) String() string {
switch {
case 0 <= i && i <= 5:
return _Class_name_0[_Class_index_0[i]:_Class_index_0[i+1]]
case i == 7:
return _Class_name_1
default:
return "Class(" + strconv.FormatInt(int64(i), 10) + ")"
}
}
package asm
import (
"fmt"
)
// Register is the source or destination of most operations.
type Register uint8
// R0 contains return values.
const R0 Register = 0
// Registers for function arguments.
const (
R1 Register = R0 + 1 + iota
R2
R3
R4
R5
)
// Callee saved registers preserved by function calls.
const (
R6 Register = R5 + 1 + iota
R7
R8
R9
)
// Read-only frame pointer to access stack.
const (
R10 Register = R9 + 1
RFP = R10
)
// Pseudo registers used by 64bit loads and jumps
const (
PseudoMapFD = R1 // BPF_PSEUDO_MAP_FD
PseudoMapValue = R2 // BPF_PSEUDO_MAP_VALUE
PseudoCall = R1 // BPF_PSEUDO_CALL
)
func (r Register) String() string {
v := uint8(r)
if v == 10 {
return "rfp"
}
return fmt.Sprintf("r%d", v)
}
package ebpf
import (
"math"
"github.com/cilium/ebpf/asm"
"github.com/cilium/ebpf/internal"
"github.com/cilium/ebpf/internal/btf"
"golang.org/x/xerrors"
)
// CollectionOptions control loading a collection into the kernel.
type CollectionOptions struct {
Programs ProgramOptions
}
// CollectionSpec describes a collection.
type CollectionSpec struct {
Maps map[string]*MapSpec
Programs map[string]*ProgramSpec
}
// Copy returns a recursive copy of the spec.
func (cs *CollectionSpec) Copy() *CollectionSpec {
if cs == nil {
return nil
}
cpy := CollectionSpec{
Maps: make(map[string]*MapSpec, len(cs.Maps)),
Programs: make(map[string]*ProgramSpec, len(cs.Programs)),
}
for name, spec := range cs.Maps {
cpy.Maps[name] = spec.Copy()
}
for name, spec := range cs.Programs {
cpy.Programs[name] = spec.Copy()
}
return &cpy
}
// RewriteMaps replaces all references to specific maps.
//
// Use this function to use pre-existing maps instead of creating new ones
// when calling NewCollection. Any named maps are removed from CollectionSpec.Maps.
//
// Returns an error if a named map isn't used in at least one program.
func (cs *CollectionSpec) RewriteMaps(maps map[string]*Map) error {
for symbol, m := range maps {
// have we seen a program that uses this symbol / map
seen := false
fd := m.FD()
for progName, progSpec := range cs.Programs {
err := progSpec.Instructions.RewriteMapPtr(symbol, fd)
switch {
case err == nil:
seen = true
case asm.IsUnreferencedSymbol(err):
// Not all programs need to use the map
default:
return xerrors.Errorf("program %s: %w", progName, err)
}
}
if !seen {
return xerrors.Errorf("map %s not referenced by any programs", symbol)
}
// Prevent NewCollection from creating rewritten maps
delete(cs.Maps, symbol)
}
return nil
}
// RewriteConstants replaces the value of multiple constants.
//
// The constant must be defined like so in the C program:
//
// static volatile const type foobar;
// static volatile const type foobar = default;
//
// Replacement values must be of the same length as the C sizeof(type).
// If necessary, they are marshalled according to the same rules as
// map values.
//
// From Linux 5.5 the verifier will use constants to eliminate dead code.
//
// Returns an error if a constant doesn't exist.
func (cs *CollectionSpec) RewriteConstants(consts map[string]interface{}) error {
rodata := cs.Maps[".rodata"]
if rodata == nil {
return xerrors.New("missing .rodata section")
}
if rodata.BTF == nil {
return xerrors.New(".rodata section has no BTF")
}
if n := len(rodata.Contents); n != 1 {
return xerrors.Errorf("expected one key in .rodata, found %d", n)
}
kv := rodata.Contents[0]
value, ok := kv.Value.([]byte)
if !ok {
return xerrors.Errorf("first value in .rodata is %T not []byte", kv.Value)
}
buf := make([]byte, len(value))
copy(buf, value)
err := patchValue(buf, btf.MapValue(rodata.BTF), consts)
if err != nil {
return err
}
rodata.Contents[0] = MapKV{kv.Key, buf}
return nil
}
// Collection is a collection of Programs and Maps associated
// with their symbols
type Collection struct {
Programs map[string]*Program
Maps map[string]*Map
}
// NewCollection creates a Collection from a specification.
//
// Only maps referenced by at least one of the programs are initialized.
func NewCollection(spec *CollectionSpec) (*Collection, error) {
return NewCollectionWithOptions(spec, CollectionOptions{})
}
// NewCollectionWithOptions creates a Collection from a specification.
//
// Only maps referenced by at least one of the programs are initialized.
func NewCollectionWithOptions(spec *CollectionSpec, opts CollectionOptions) (coll *Collection, err error) {
var (
maps = make(map[string]*Map)
progs = make(map[string]*Program)
btfs = make(map[*btf.Spec]*btf.Handle)
)
defer func() {
for _, btf := range btfs {
btf.Close()
}
if err == nil {
return
}
for _, m := range maps {
m.Close()
}
for _, p := range progs {
p.Close()
}
}()
loadBTF := func(spec *btf.Spec) (*btf.Handle, error) {
if btfs[spec] != nil {
return btfs[spec], nil
}
handle, err := btf.NewHandle(spec)
if err != nil {
return nil, err
}
btfs[spec] = handle
return handle, nil
}
for mapName, mapSpec := range spec.Maps {
var handle *btf.Handle
if mapSpec.BTF != nil {
handle, err = loadBTF(btf.MapSpec(mapSpec.BTF))
if err != nil && !xerrors.Is(err, btf.ErrNotSupported) {
return nil, err
}
}
m, err := newMapWithBTF(mapSpec, handle)
if err != nil {
return nil, xerrors.Errorf("map %s: %w", mapName, err)
}
maps[mapName] = m
}
for progName, origProgSpec := range spec.Programs {
progSpec := origProgSpec.Copy()
// Rewrite any reference to a valid map.
for i := range progSpec.Instructions {
ins := &progSpec.Instructions[i]
if ins.OpCode != asm.LoadImmOp(asm.DWord) || ins.Reference == "" {
continue
}
if uint32(ins.Constant) != math.MaxUint32 {
// Don't overwrite maps already rewritten, users can
// rewrite programs in the spec themselves
continue
}
m := maps[ins.Reference]
if m == nil {
return nil, xerrors.Errorf("program %s: missing map %s", progName, ins.Reference)
}
fd := m.FD()
if fd < 0 {
return nil, xerrors.Errorf("map %s: %w", ins.Reference, internal.ErrClosedFd)
}
if err := ins.RewriteMapPtr(m.FD()); err != nil {
return nil, xerrors.Errorf("progam %s: map %s: %w", progName, ins.Reference, err)
}
}
var handle *btf.Handle
if progSpec.BTF != nil {
handle, err = loadBTF(btf.ProgramSpec(progSpec.BTF))
if err != nil && !xerrors.Is(err, btf.ErrNotSupported) {
return nil, err
}
}
prog, err := newProgramWithBTF(progSpec, handle, opts.Programs)
if err != nil {
return nil, xerrors.Errorf("program %s: %w", progName, err)
}
progs[progName] = prog
}
return &Collection{
progs,
maps,
}, nil
}
// LoadCollection parses an object file and converts it to a collection.
func LoadCollection(file string) (*Collection, error) {
spec, err := LoadCollectionSpec(file)
if err != nil {
return nil, err
}
return NewCollection(spec)
}
// Close frees all maps and programs associated with the collection.
//
// The collection mustn't be used afterwards.
func (coll *Collection) Close() {
for _, prog := range coll.Programs {
prog.Close()
}
for _, m := range coll.Maps {
m.Close()
}
}
// DetachMap removes the named map from the Collection.
//
// This means that a later call to Close() will not affect this map.
//
// Returns nil if no map of that name exists.
func (coll *Collection) DetachMap(name string) *Map {
m := coll.Maps[name]
delete(coll.Maps, name)
return m
}
// DetachProgram removes the named program from the Collection.
//
// This means that a later call to Close() will not affect this program.
//
// Returns nil if no program of that name exists.
func (coll *Collection) DetachProgram(name string) *Program {
p := coll.Programs[name]
delete(coll.Programs, name)
return p
}
// Package ebpf is a toolkit for working with eBPF programs.
//
// eBPF programs are small snippets of code which are executed directly
// in a VM in the Linux kernel, which makes them very fast and flexible.
// Many Linux subsystems now accept eBPF programs. This makes it possible
// to implement highly application specific logic inside the kernel,
// without having to modify the actual kernel itself.
//
// This package is designed for long-running processes which
// want to use eBPF to implement part of their application logic. It has no
// run-time dependencies outside of the library and the Linux kernel itself.
// eBPF code should be compiled ahead of time using clang, and shipped with
// your application as any other resource.
//
// This package doesn't include code required to attach eBPF to Linux
// subsystems, since this varies per subsystem.
package ebpf
module github.com/cilium/ebpf
go 1.12
require (
golang.org/x/sys v0.0.0-20200124204421-9fbb57f87de9
golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543
)
golang.org/x/sys v0.0.0-20191022100944-742c48ecaeb7 h1:HmbHVPwrPEKPGLAcHSrMe6+hqSUlvZU0rab6x5EXfGU=
golang.org/x/sys v0.0.0-20191022100944-742c48ecaeb7/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
golang.org/x/sys v0.0.0-20200124204421-9fbb57f87de9 h1:1/DFK4b7JH8DmkqhUk48onnSfrPzImPoVxuomtbT2nk=
golang.org/x/sys v0.0.0-20200124204421-9fbb57f87de9/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543 h1:E7g+9GITq07hpfrRu66IVDexMakfv52eLZ2CXBWiKr4=
golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
package btf
import (
"encoding/binary"
"fmt"
"io"
"golang.org/x/xerrors"
)
// btfKind describes a Type.
type btfKind uint8
// Equivalents of the BTF_KIND_* constants.
const (
kindUnknown btfKind = iota
kindInt
kindPointer
kindArray
kindStruct
kindUnion
kindEnum
kindForward
kindTypedef
kindVolatile
kindConst
kindRestrict
// Added ~4.20
kindFunc
kindFuncProto
// Added ~5.1
kindVar
kindDatasec
)
const (
btfTypeKindShift = 24
btfTypeKindLen = 4
btfTypeVlenShift = 0
btfTypeVlenMask = 16
)
// btfType is equivalent to struct btf_type in Documentation/bpf/btf.rst.
type btfType struct {
NameOff uint32
/* "info" bits arrangement
* bits 0-15: vlen (e.g. # of struct's members)
* bits 16-23: unused
* bits 24-27: kind (e.g. int, ptr, array...etc)
* bits 28-30: unused
* bit 31: kind_flag, currently used by
* struct, union and fwd
*/
Info uint32
/* "size" is used by INT, ENUM, STRUCT and UNION.
* "size" tells the size of the type it is describing.
*
* "type" is used by PTR, TYPEDEF, VOLATILE, CONST, RESTRICT,
* FUNC and FUNC_PROTO.
* "type" is a type_id referring to another type.
*/
SizeType uint32
}
func (k btfKind) String() string {
switch k {
case kindUnknown:
return "Unknown"
case kindInt:
return "Integer"
case kindPointer:
return "Pointer"
case kindArray:
return "Array"
case kindStruct:
return "Struct"
case kindUnion:
return "Union"
case kindEnum:
return "Enumeration"
case kindForward:
return "Forward"
case kindTypedef:
return "Typedef"
case kindVolatile:
return "Volatile"
case kindConst:
return "Const"
case kindRestrict:
return "Restrict"
case kindFunc:
return "Function"
case kindFuncProto:
return "Function Proto"
case kindVar:
return "Variable"
case kindDatasec:
return "Section"
default:
return fmt.Sprintf("Unknown (%d)", k)
}
}
func mask(len uint32) uint32 {
return (1 << len) - 1
}
func (bt *btfType) info(len, shift uint32) uint32 {
return (bt.Info >> shift) & mask(len)
}
func (bt *btfType) setInfo(value, len, shift uint32) {
bt.Info &^= mask(len) << shift
bt.Info |= (value & mask(len)) << shift
}
func (bt *btfType) Kind() btfKind {
return btfKind(bt.info(btfTypeKindLen, btfTypeKindShift))
}
func (bt *btfType) SetKind(kind btfKind) {
bt.setInfo(uint32(kind), btfTypeKindLen, btfTypeKindShift)
}
func (bt *btfType) Vlen() int {
return int(bt.info(btfTypeVlenMask, btfTypeVlenShift))
}
func (bt *btfType) SetVlen(vlen int) {
bt.setInfo(uint32(vlen), btfTypeVlenMask, btfTypeVlenShift)
}
func (bt *btfType) Type() TypeID {
// TODO: Panic here if wrong kind?
return TypeID(bt.SizeType)
}
func (bt *btfType) Size() uint32 {
// TODO: Panic here if wrong kind?
return bt.SizeType
}
type rawType struct {
btfType
data interface{}
}
func (rt *rawType) Marshal(w io.Writer, bo binary.ByteOrder) error {
if err := binary.Write(w, bo, &rt.btfType); err != nil {
return err
}
if rt.data == nil {
return nil
}
return binary.Write(w, bo, rt.data)
}
type btfArray struct {
Type TypeID
IndexType TypeID
Nelems uint32
}
type btfMember struct {
NameOff uint32
Type TypeID
Offset uint32
}
type btfVarSecinfo struct {
Type TypeID
Offset uint32
Size uint32
}
type btfVariable struct {
Linkage uint32
}
func readTypes(r io.Reader, bo binary.ByteOrder) ([]rawType, error) {
var (
header btfType
types []rawType
)
for id := TypeID(1); ; id++ {
if err := binary.Read(r, bo, &header); err == io.EOF {
return types, nil
} else if err != nil {
return nil, xerrors.Errorf("can't read type info for id %v: %v", id, err)
}
var data interface{}
switch header.Kind() {
case kindInt:
// sizeof(uint32)
data = make([]byte, 4)
case kindPointer:
case kindArray:
data = new(btfArray)
case kindStruct:
fallthrough
case kindUnion:
data = make([]btfMember, header.Vlen())
case kindEnum:
// sizeof(struct btf_enum)
data = make([]byte, header.Vlen()*4*2)
case kindForward:
case kindTypedef:
case kindVolatile:
case kindConst:
case kindRestrict:
case kindFunc:
case kindFuncProto:
// sizeof(struct btf_param)
data = make([]byte, header.Vlen()*4*2)
case kindVar:
data = new(btfVariable)
case kindDatasec:
data = make([]btfVarSecinfo, header.Vlen())
default:
return nil, xerrors.Errorf("type id %v: unknown kind: %v", id, header.Kind())
}
if data == nil {
types = append(types, rawType{header, nil})
continue
}
if err := binary.Read(r, bo, data); err != nil {
return nil, xerrors.Errorf("type id %d: kind %v: can't read %T: %v", id, header.Kind(), data, err)
}
types = append(types, rawType{header, data})
}
}
// Package btf handles data encoded according to the BPF Type Format.
//
// The canonical documentation lives in the Linux kernel repository and is
// available at https://www.kernel.org/doc/html/latest/bpf/btf.html
//
// The API is very much unstable. You should only use this via the main
// ebpf library.
package btf
package btf
import (
"bytes"
"encoding/binary"
"io"
"io/ioutil"
"github.com/cilium/ebpf/asm"
"github.com/cilium/ebpf/internal"
"golang.org/x/xerrors"
)
type btfExtHeader struct {
Magic uint16
Version uint8
Flags uint8
HdrLen uint32
FuncInfoOff uint32
FuncInfoLen uint32
LineInfoOff uint32
LineInfoLen uint32
}
func parseExtInfos(r io.ReadSeeker, bo binary.ByteOrder, strings stringTable) (funcInfo, lineInfo map[string]extInfo, err error) {
var header btfExtHeader
if err := binary.Read(r, bo, &header); err != nil {
return nil, nil, xerrors.Errorf("can't read header: %v", err)
}
if header.Magic != btfMagic {
return nil, nil, xerrors.Errorf("incorrect magic value %v", header.Magic)
}
if header.Version != 1 {
return nil, nil, xerrors.Errorf("unexpected version %v", header.Version)
}
if header.Flags != 0 {
return nil, nil, xerrors.Errorf("unsupported flags %v", header.Flags)
}
remainder := int64(header.HdrLen) - int64(binary.Size(&header))
if remainder < 0 {
return nil, nil, xerrors.New("header is too short")
}
// Of course, the .BTF.ext header has different semantics than the
// .BTF ext header. We need to ignore non-null values.
_, err = io.CopyN(ioutil.Discard, r, remainder)
if err != nil {
return nil, nil, xerrors.Errorf("header padding: %v", err)
}
if _, err := r.Seek(int64(header.HdrLen+header.FuncInfoOff), io.SeekStart); err != nil {
return nil, nil, xerrors.Errorf("can't seek to function info section: %v", err)
}
funcInfo, err = parseExtInfo(io.LimitReader(r, int64(header.FuncInfoLen)), bo, strings)
if err != nil {
return nil, nil, xerrors.Errorf("function info: %w", err)
}
if _, err := r.Seek(int64(header.HdrLen+header.LineInfoOff), io.SeekStart); err != nil {
return nil, nil, xerrors.Errorf("can't seek to line info section: %v", err)
}
lineInfo, err = parseExtInfo(io.LimitReader(r, int64(header.LineInfoLen)), bo, strings)
if err != nil {
return nil, nil, xerrors.Errorf("line info: %w", err)
}
return funcInfo, lineInfo, nil
}
type btfExtInfoSec struct {
SecNameOff uint32
NumInfo uint32
}
type extInfoRecord struct {
InsnOff uint64
Opaque []byte
}
type extInfo struct {
recordSize uint32
records []extInfoRecord
}
func (ei extInfo) append(other extInfo, offset uint64) (extInfo, error) {
if other.recordSize != ei.recordSize {
return extInfo{}, xerrors.Errorf("ext_info record size mismatch, want %d (got %d)", ei.recordSize, other.recordSize)
}
records := make([]extInfoRecord, 0, len(ei.records)+len(other.records))
records = append(records, ei.records...)
for _, info := range other.records {
records = append(records, extInfoRecord{
InsnOff: info.InsnOff + offset,
Opaque: info.Opaque,
})
}
return extInfo{ei.recordSize, records}, nil
}
func (ei extInfo) MarshalBinary() ([]byte, error) {
if len(ei.records) == 0 {
return nil, nil
}
buf := bytes.NewBuffer(make([]byte, 0, int(ei.recordSize)*len(ei.records)))
for _, info := range ei.records {
// The kernel expects offsets in number of raw bpf instructions,
// while the ELF tracks it in bytes.
insnOff := uint32(info.InsnOff / asm.InstructionSize)
if err := binary.Write(buf, internal.NativeEndian, insnOff); err != nil {
return nil, xerrors.Errorf("can't write instruction offset: %v", err)
}
buf.Write(info.Opaque)
}
return buf.Bytes(), nil
}
func parseExtInfo(r io.Reader, bo binary.ByteOrder, strings stringTable) (map[string]extInfo, error) {
var recordSize uint32
if err := binary.Read(r, bo, &recordSize); err != nil {
return nil, xerrors.Errorf("can't read record size: %v", err)
}
if recordSize < 4 {
// Need at least insnOff
return nil, xerrors.New("record size too short")
}
result := make(map[string]extInfo)
for {
var infoHeader btfExtInfoSec
if err := binary.Read(r, bo, &infoHeader); err == io.EOF {
return result, nil
} else if err != nil {
return nil, xerrors.Errorf("can't read ext info header: %v", err)
}
secName, err := strings.Lookup(infoHeader.SecNameOff)
if err != nil {
return nil, xerrors.Errorf("can't get section name: %w", err)
}
if infoHeader.NumInfo == 0 {
return nil, xerrors.Errorf("section %s has invalid number of records", secName)
}
var records []extInfoRecord
for i := uint32(0); i < infoHeader.NumInfo; i++ {
var byteOff uint32
if err := binary.Read(r, bo, &byteOff); err != nil {
return nil, xerrors.Errorf("section %v: can't read extended info offset: %v", secName, err)
}
buf := make([]byte, int(recordSize-4))
if _, err := io.ReadFull(r, buf); err != nil {
return nil, xerrors.Errorf("section %v: can't read record: %v", secName, err)
}
if byteOff%asm.InstructionSize != 0 {
return nil, xerrors.Errorf("section %v: offset %v is not aligned with instruction size", secName, byteOff)
}
records = append(records, extInfoRecord{uint64(byteOff), buf})
}
result[secName] = extInfo{
recordSize,
records,
}
}
}
package btf
import (
"bytes"
"io"
"io/ioutil"
"golang.org/x/xerrors"
)
type stringTable []byte
func readStringTable(r io.Reader) (stringTable, error) {
contents, err := ioutil.ReadAll(r)
if err != nil {
return nil, xerrors.Errorf("can't read string table: %v", err)
}
if len(contents) < 1 {
return nil, xerrors.New("string table is empty")
}
if contents[0] != '\x00' {
return nil, xerrors.New("first item in string table is non-empty")
}
if contents[len(contents)-1] != '\x00' {
return nil, xerrors.New("string table isn't null terminated")
}
return stringTable(contents), nil
}
func (st stringTable) Lookup(offset uint32) (string, error) {
if int64(offset) > int64(^uint(0)>>1) {
return "", xerrors.Errorf("offset %d overflows int", offset)
}
pos := int(offset)
if pos >= len(st) {
return "", xerrors.Errorf("offset %d is out of bounds", offset)
}
if pos > 0 && st[pos-1] != '\x00' {
return "", xerrors.Errorf("offset %d isn't start of a string", offset)
}
str := st[pos:]
end := bytes.IndexByte(str, '\x00')
if end == -1 {
return "", xerrors.Errorf("offset %d isn't null terminated", offset)
}
return string(str[:end]), nil
}
func (st stringTable) LookupName(offset uint32) (Name, error) {
str, err := st.Lookup(offset)
return Name(str), err
}
package internal
import (
"fmt"
"io/ioutil"
"strings"
"sync"
)
var sysCPU struct {
once sync.Once
err error
num int
}
// PossibleCPUs returns the max number of CPUs a system may possibly have
// Logical CPU numbers must be of the form 0-n
func PossibleCPUs() (int, error) {
sysCPU.once.Do(func() {
sysCPU.num, sysCPU.err = parseCPUsFromFile("/sys/devices/system/cpu/possible")
})
return sysCPU.num, sysCPU.err
}
func parseCPUsFromFile(path string) (int, error) {
spec, err := ioutil.ReadFile(path)
if err != nil {
return 0, err
}
n, err := parseCPUs(string(spec))
if err != nil {
return 0, fmt.Errorf("can't parse %s: %v", path, err)
}
return n, nil
}
// parseCPUs parses the number of cpus from a string produced
// by bitmap_list_string() in the Linux kernel.
// Multiple ranges are rejected, since they can't be unified
// into a single number.
// This is the format of /sys/devices/system/cpu/possible, it
// is not suitable for /sys/devices/system/cpu/online, etc.
func parseCPUs(spec string) (int, error) {
if strings.Trim(spec, "\n") == "0" {
return 1, nil
}
var low, high int
n, err := fmt.Sscanf(spec, "%d-%d\n", &low, &high)
if n != 2 || err != nil {
return 0, fmt.Errorf("invalid format: %s", spec)
}
if low != 0 {
return 0, fmt.Errorf("CPU spec doesn't start at zero: %s", spec)
}
// cpus is 0 indexed
return high + 1, nil
}
package internal
import (
"encoding/binary"
"unsafe"
)
// NativeEndian is set to either binary.BigEndian or binary.LittleEndian,
// depending on the host's endianness.
var NativeEndian binary.ByteOrder
func init() {
if isBigEndian() {
NativeEndian = binary.BigEndian
} else {
NativeEndian = binary.LittleEndian
}
}
func isBigEndian() (ret bool) {
i := int(0x1)
bs := (*[int(unsafe.Sizeof(i))]byte)(unsafe.Pointer(&i))
return bs[0] == 0
}
package internal
import (
"bytes"
"fmt"
"strings"
"github.com/cilium/ebpf/internal/unix"
"golang.org/x/xerrors"
)
// ErrorWithLog returns an error that includes logs from the
// kernel verifier.
//
// logErr should be the error returned by the syscall that generated
// the log. It is used to check for truncation of the output.
func ErrorWithLog(err error, log []byte, logErr error) error {
logStr := strings.Trim(CString(log), "\t\r\n ")
if xerrors.Is(logErr, unix.ENOSPC) {
logStr += " (truncated...)"
}
return &VerifierError{err, logStr}
}
// VerifierError includes information from the eBPF verifier.
type VerifierError struct {
cause error
log string
}
func (le *VerifierError) Error() string {
if le.log == "" {
return le.cause.Error()
}
return fmt.Sprintf("%s: %s", le.cause, le.log)
}
// CString turns a NUL / zero terminated byte buffer into a string.
func CString(in []byte) string {
inLen := bytes.IndexByte(in, 0)
if inLen == -1 {
return ""
}
return string(in[:inLen])
}
package internal
import (
"runtime"
"strconv"
"github.com/cilium/ebpf/internal/unix"
"golang.org/x/xerrors"
)
var ErrClosedFd = xerrors.New("use of closed file descriptor")
type FD struct {
raw int64
}
func NewFD(value uint32) *FD {
fd := &FD{int64(value)}
runtime.SetFinalizer(fd, (*FD).Close)
return fd
}
func (fd *FD) String() string {
return strconv.FormatInt(fd.raw, 10)
}
func (fd *FD) Value() (uint32, error) {
if fd.raw < 0 {
return 0, ErrClosedFd
}
return uint32(fd.raw), nil
}
func (fd *FD) Close() error {
if fd.raw < 0 {
return nil
}
value := int(fd.raw)
fd.raw = -1
fd.Forget()
return unix.Close(value)
}
func (fd *FD) Forget() {
runtime.SetFinalizer(fd, nil)
}
func (fd *FD) Dup() (*FD, error) {
if fd.raw < 0 {
return nil, ErrClosedFd
}
dup, err := unix.FcntlInt(uintptr(fd.raw), unix.F_DUPFD_CLOEXEC, 0)
if err != nil {
return nil, xerrors.Errorf("can't dup fd: %v", err)
}
return NewFD(uint32(dup)), nil
}
package internal
import (
"fmt"
"sync"
"golang.org/x/xerrors"
)
// ErrNotSupported indicates that a feature is not supported by the current kernel.
var ErrNotSupported = xerrors.New("not supported")
// UnsupportedFeatureError is returned by FeatureTest() functions.
type UnsupportedFeatureError struct {
// The minimum Linux mainline version required for this feature.
// Used for the error string, and for sanity checking during testing.
MinimumVersion Version
// The name of the feature that isn't supported.
Name string
}
func (ufe *UnsupportedFeatureError) Error() string {
return fmt.Sprintf("%s not supported (requires >= %s)", ufe.Name, ufe.MinimumVersion)
}
// Is indicates that UnsupportedFeatureError is ErrNotSupported.
func (ufe *UnsupportedFeatureError) Is(target error) bool {
return target == ErrNotSupported
}
// FeatureTest wraps a function so that it is run at most once.
//
// name should identify the tested feature, while version must be in the
// form Major.Minor[.Patch].
//
// Returns a descriptive UnsupportedFeatureError if the feature is not available.
func FeatureTest(name, version string, fn func() bool) func() error {
v, err := NewVersion(version)
if err != nil {
return func() error { return err }
}
var (
once sync.Once
result error
)
return func() error {
once.Do(func() {
if !fn() {
result = &UnsupportedFeatureError{
MinimumVersion: v,
Name: name,
}
}
})
return result
}
}
// A Version in the form Major.Minor.Patch.
type Version [3]uint16
// NewVersion creates a version from a string like "Major.Minor.Patch".
//
// Patch is optional.
func NewVersion(ver string) (Version, error) {
var major, minor, patch uint16
n, _ := fmt.Sscanf(ver, "%d.%d.%d", &major, &minor, &patch)
if n < 2 {
return Version{}, xerrors.Errorf("invalid version: %s", ver)
}
return Version{major, minor, patch}, nil
}
func (v Version) String() string {
if v[2] == 0 {
return fmt.Sprintf("v%d.%d", v[0], v[1])
}
return fmt.Sprintf("v%d.%d.%d", v[0], v[1], v[2])
}
// Less returns true if the version is less than another version.
func (v Version) Less(other Version) bool {
for i, a := range v {
if a == other[i] {
continue
}
return a < other[i]
}
return false
}
package internal
import "golang.org/x/xerrors"
// DiscardZeroes makes sure that all written bytes are zero
// before discarding them.
type DiscardZeroes struct{}
func (DiscardZeroes) Write(p []byte) (int, error) {
for _, b := range p {
if b != 0 {
return 0, xerrors.New("encountered non-zero byte")
}
}
return len(p), nil
}
package internal
import "unsafe"
// NewPointer creates a 64-bit pointer from an unsafe Pointer.
func NewPointer(ptr unsafe.Pointer) Pointer {
return Pointer{ptr: ptr}
}
// NewSlicePointer creates a 64-bit pointer from a byte slice.
func NewSlicePointer(buf []byte) Pointer {
if len(buf) == 0 {
return Pointer{}
}
return Pointer{ptr: unsafe.Pointer(&buf[0])}
}
// NewStringPointer creates a 64-bit pointer from a string.
func NewStringPointer(str string) Pointer {
if str == "" {
return Pointer{}
}
// The kernel expects strings to be zero terminated
buf := make([]byte, len(str)+1)
copy(buf, str)
return Pointer{ptr: unsafe.Pointer(&buf[0])}
}
// +build armbe mips mips64p32
package internal
import (
"unsafe"
)
// Pointer wraps an unsafe.Pointer to be 64bit to
// conform to the syscall specification.
type Pointer struct {
pad uint32
ptr unsafe.Pointer
}
// +build 386 amd64p32 arm mipsle mips64p32le
package internal
import (
"unsafe"
)
// Pointer wraps an unsafe.Pointer to be 64bit to
// conform to the syscall specification.
type Pointer struct {
ptr unsafe.Pointer
pad uint32
}
// +build !386,!amd64p32,!arm,!mipsle,!mips64p32le
// +build !armbe,!mips,!mips64p32
package internal
import (
"unsafe"
)
// Pointer wraps an unsafe.Pointer to be 64bit to
// conform to the syscall specification.
type Pointer struct {
ptr unsafe.Pointer
}
package internal
import (
"runtime"
"unsafe"
"github.com/cilium/ebpf/internal/unix"
)
// BPF wraps SYS_BPF.
//
// Any pointers contained in attr must use the Pointer type from this package.
func BPF(cmd int, attr unsafe.Pointer, size uintptr) (uintptr, error) {
r1, _, errNo := unix.Syscall(unix.SYS_BPF, uintptr(cmd), uintptr(attr), size)
runtime.KeepAlive(attr)
var err error
if errNo != 0 {
err = errNo
}
return r1, err
}
// +build linux
package unix
import (
"syscall"
linux "golang.org/x/sys/unix"
)
const (
ENOENT = linux.ENOENT
EAGAIN = linux.EAGAIN
ENOSPC = linux.ENOSPC
EINVAL = linux.EINVAL
EPOLLIN = linux.EPOLLIN
EINTR = linux.EINTR
ESRCH = linux.ESRCH
ENODEV = linux.ENODEV
BPF_F_RDONLY_PROG = linux.BPF_F_RDONLY_PROG
BPF_F_WRONLY_PROG = linux.BPF_F_WRONLY_PROG
BPF_OBJ_NAME_LEN = linux.BPF_OBJ_NAME_LEN
BPF_TAG_SIZE = linux.BPF_TAG_SIZE
SYS_BPF = linux.SYS_BPF
F_DUPFD_CLOEXEC = linux.F_DUPFD_CLOEXEC
EPOLL_CTL_ADD = linux.EPOLL_CTL_ADD
EPOLL_CLOEXEC = linux.EPOLL_CLOEXEC
O_CLOEXEC = linux.O_CLOEXEC
O_NONBLOCK = linux.O_NONBLOCK
PROT_READ = linux.PROT_READ
PROT_WRITE = linux.PROT_WRITE
MAP_SHARED = linux.MAP_SHARED
PERF_TYPE_SOFTWARE = linux.PERF_TYPE_SOFTWARE
PERF_COUNT_SW_BPF_OUTPUT = linux.PERF_COUNT_SW_BPF_OUTPUT
PerfBitWatermark = linux.PerfBitWatermark
PERF_SAMPLE_RAW = linux.PERF_SAMPLE_RAW
PERF_FLAG_FD_CLOEXEC = linux.PERF_FLAG_FD_CLOEXEC
RLIM_INFINITY = linux.RLIM_INFINITY
RLIMIT_MEMLOCK = linux.RLIMIT_MEMLOCK
)
// Statfs_t is a wrapper
type Statfs_t = linux.Statfs_t
// Rlimit is a wrapper
type Rlimit = linux.Rlimit
// Setrlimit is a wrapper
func Setrlimit(resource int, rlim *Rlimit) (err error) {
return linux.Setrlimit(resource, rlim)
}
// Syscall is a wrapper
func Syscall(trap, a1, a2, a3 uintptr) (r1, r2 uintptr, err syscall.Errno) {
return linux.Syscall(trap, a1, a2, a3)
}
// FcntlInt is a wrapper
func FcntlInt(fd uintptr, cmd, arg int) (int, error) {
return linux.FcntlInt(fd, cmd, arg)
}
// Statfs is a wrapper
func Statfs(path string, buf *Statfs_t) (err error) {
return linux.Statfs(path, buf)
}
// Close is a wrapper
func Close(fd int) (err error) {
return linux.Close(fd)
}
// EpollEvent is a wrapper
type EpollEvent = linux.EpollEvent
// EpollWait is a wrapper
func EpollWait(epfd int, events []EpollEvent, msec int) (n int, err error) {
return linux.EpollWait(epfd, events, msec)
}
// EpollCtl is a wrapper
func EpollCtl(epfd int, op int, fd int, event *EpollEvent) (err error) {
return linux.EpollCtl(epfd, op, fd, event)
}
// Eventfd is a wrapper
func Eventfd(initval uint, flags int) (fd int, err error) {
return linux.Eventfd(initval, flags)
}
// Write is a wrapper
func Write(fd int, p []byte) (n int, err error) {
return linux.Write(fd, p)
}
// EpollCreate1 is a wrapper
func EpollCreate1(flag int) (fd int, err error) {
return linux.EpollCreate1(flag)
}
// PerfEventMmapPage is a wrapper
type PerfEventMmapPage linux.PerfEventMmapPage
// SetNonblock is a wrapper
func SetNonblock(fd int, nonblocking bool) (err error) {
return linux.SetNonblock(fd, nonblocking)
}
// Mmap is a wrapper
func Mmap(fd int, offset int64, length int, prot int, flags int) (data []byte, err error) {
return linux.Mmap(fd, offset, length, prot, flags)
}
// Munmap is a wrapper
func Munmap(b []byte) (err error) {
return linux.Munmap(b)
}
// PerfEventAttr is a wrapper
type PerfEventAttr = linux.PerfEventAttr
// PerfEventOpen is a wrapper
func PerfEventOpen(attr *PerfEventAttr, pid int, cpu int, groupFd int, flags int) (fd int, err error) {
return linux.PerfEventOpen(attr, pid, cpu, groupFd, flags)
}
// Utsname is a wrapper
type Utsname = linux.Utsname
// Uname is a wrapper
func Uname(buf *Utsname) (err error) {
return linux.Uname(buf)
}
// Getpid is a wrapper
func Getpid() int {
return linux.Getpid()
}
// Gettid is a wrapper
func Gettid() int {
return linux.Gettid()
}
// Tgkill is a wrapper
func Tgkill(tgid int, tid int, sig syscall.Signal) (err error) {
return linux.Tgkill(tgid, tid, sig)
}
// +build !linux
package unix
import (
"fmt"
"runtime"
"syscall"
)
var errNonLinux = fmt.Errorf("unsupported platform %s/%s", runtime.GOOS, runtime.GOARCH)
const (
ENOENT = syscall.ENOENT
EAGAIN = syscall.EAGAIN
ENOSPC = syscall.ENOSPC
EINVAL = syscall.EINVAL
EINTR = syscall.EINTR
ESRCH = syscall.ESRCH
ENODEV = syscall.ENODEV
BPF_F_RDONLY_PROG = 0
BPF_F_WRONLY_PROG = 0
BPF_OBJ_NAME_LEN = 0x10
BPF_TAG_SIZE = 0x8
SYS_BPF = 321
F_DUPFD_CLOEXEC = 0x406
EPOLLIN = 0x1
EPOLL_CTL_ADD = 0x1
EPOLL_CLOEXEC = 0x80000
O_CLOEXEC = 0x80000
O_NONBLOCK = 0x800
PROT_READ = 0x1
PROT_WRITE = 0x2
MAP_SHARED = 0x1
PERF_TYPE_SOFTWARE = 0x1
PERF_COUNT_SW_BPF_OUTPUT = 0xa
PerfBitWatermark = 0x4000
PERF_SAMPLE_RAW = 0x400
PERF_FLAG_FD_CLOEXEC = 0x8
RLIM_INFINITY = 0x7fffffffffffffff
RLIMIT_MEMLOCK = 8
)
// Statfs_t is a wrapper
type Statfs_t struct {
Type int64
Bsize int64
Blocks uint64
Bfree uint64
Bavail uint64
Files uint64
Ffree uint64
Fsid [2]int32
Namelen int64
Frsize int64
Flags int64
Spare [4]int64
}
// Rlimit is a wrapper
type Rlimit struct {
Cur uint64
Max uint64
}
// Setrlimit is a wrapper
func Setrlimit(resource int, rlim *Rlimit) (err error) {
return errNonLinux
}
// Syscall is a wrapper
func Syscall(trap, a1, a2, a3 uintptr) (r1, r2 uintptr, err syscall.Errno) {
return 0, 0, syscall.Errno(1)
}
// FcntlInt is a wrapper
func FcntlInt(fd uintptr, cmd, arg int) (int, error) {
return -1, errNonLinux
}
// Statfs is a wrapper
func Statfs(path string, buf *Statfs_t) error {
return errNonLinux
}
// Close is a wrapper
func Close(fd int) (err error) {
return errNonLinux
}
// EpollEvent is a wrapper
type EpollEvent struct {
Events uint32
Fd int32
Pad int32
}
// EpollWait is a wrapper
func EpollWait(epfd int, events []EpollEvent, msec int) (n int, err error) {
return 0, errNonLinux
}
// EpollCtl is a wrapper
func EpollCtl(epfd int, op int, fd int, event *EpollEvent) (err error) {
return errNonLinux
}
// Eventfd is a wrapper
func Eventfd(initval uint, flags int) (fd int, err error) {
return 0, errNonLinux
}
// Write is a wrapper
func Write(fd int, p []byte) (n int, err error) {
return 0, errNonLinux
}
// EpollCreate1 is a wrapper
func EpollCreate1(flag int) (fd int, err error) {
return 0, errNonLinux
}
// PerfEventMmapPage is a wrapper
type PerfEventMmapPage struct {
Version uint32
Compat_version uint32
Lock uint32
Index uint32
Offset int64
Time_enabled uint64
Time_running uint64
Capabilities uint64
Pmc_width uint16
Time_shift uint16
Time_mult uint32
Time_offset uint64
Time_zero uint64
Size uint32
Data_head uint64
Data_tail uint64
Data_offset uint64
Data_size uint64
Aux_head uint64
Aux_tail uint64
Aux_offset uint64
Aux_size uint64
}
// SetNonblock is a wrapper
func SetNonblock(fd int, nonblocking bool) (err error) {
return errNonLinux
}
// Mmap is a wrapper
func Mmap(fd int, offset int64, length int, prot int, flags int) (data []byte, err error) {
return []byte{}, errNonLinux
}
// Munmap is a wrapper
func Munmap(b []byte) (err error) {
return errNonLinux
}
// PerfEventAttr is a wrapper
type PerfEventAttr struct {
Type uint32
Size uint32
Config uint64
Sample uint64
Sample_type uint64
Read_format uint64
Bits uint64
Wakeup uint32
Bp_type uint32
Ext1 uint64
Ext2 uint64
Branch_sample_type uint64
Sample_regs_user uint64
Sample_stack_user uint32
Clockid int32
Sample_regs_intr uint64
Aux_watermark uint32
Sample_max_stack uint16
}
// PerfEventOpen is a wrapper
func PerfEventOpen(attr *PerfEventAttr, pid int, cpu int, groupFd int, flags int) (fd int, err error) {
return 0, errNonLinux
}
// Utsname is a wrapper
type Utsname struct {
Release [65]byte
}
// Uname is a wrapper
func Uname(buf *Utsname) (err error) {
return errNonLinux
}
// Getpid is a wrapper
func Getpid() int {
return -1
}
// Gettid is a wrapper
func Gettid() int {
return -1
}
// Tgkill is a wrapper
func Tgkill(tgid int, tid int, sig syscall.Signal) (err error) {
return errNonLinux
}
package ebpf
import (
"github.com/cilium/ebpf/asm"
"github.com/cilium/ebpf/internal/btf"
"golang.org/x/xerrors"
)
// link resolves bpf-to-bpf calls.
//
// Each library may contain multiple functions / labels, and is only linked
// if prog references one of these functions.
//
// Libraries also linked.
func link(prog *ProgramSpec, libs []*ProgramSpec) error {
var (
linked = make(map[*ProgramSpec]bool)
pending = []asm.Instructions{prog.Instructions}
insns asm.Instructions
)
for len(pending) > 0 {
insns, pending = pending[0], pending[1:]
for _, lib := range libs {
if linked[lib] {
continue
}
needed, err := needSection(insns, lib.Instructions)
if err != nil {
return xerrors.Errorf("linking %s: %w", lib.Name, err)
}
if !needed {
continue
}
linked[lib] = true
prog.Instructions = append(prog.Instructions, lib.Instructions...)
pending = append(pending, lib.Instructions)
if prog.BTF != nil && lib.BTF != nil {
if err := btf.ProgramAppend(prog.BTF, lib.BTF); err != nil {
return xerrors.Errorf("linking BTF of %s: %w", lib.Name, err)
}
}
}
}
return nil
}
func needSection(insns, section asm.Instructions) (bool, error) {
// A map of symbols to the libraries which contain them.
symbols, err := section.SymbolOffsets()
if err != nil {
return false, err
}
for _, ins := range insns {
if ins.Reference == "" {
continue
}
if ins.OpCode.JumpOp() != asm.Call || ins.Src != asm.PseudoCall {
continue
}
if ins.Constant != -1 {
// This is already a valid call, no need to link again.
continue
}
if _, ok := symbols[ins.Reference]; !ok {
// Symbol isn't available in this section
continue
}
// At this point we know that at least one function in the
// library is called from insns, so we have to link it.
return true, nil
}
// None of the functions in the section are called.
return false, nil
}
package ebpf
import (
"bytes"
"encoding"
"encoding/binary"
"reflect"
"runtime"
"unsafe"
"github.com/cilium/ebpf/internal"
"golang.org/x/xerrors"
)
func marshalPtr(data interface{}, length int) (internal.Pointer, error) {
if data == nil {
if length == 0 {
return internal.NewPointer(nil), nil
}
return internal.Pointer{}, xerrors.New("can't use nil as key of map")
}
if ptr, ok := data.(unsafe.Pointer); ok {
return internal.NewPointer(ptr), nil
}
buf, err := marshalBytes(data, length)
if err != nil {
return internal.Pointer{}, err
}
return internal.NewSlicePointer(buf), nil
}
func marshalBytes(data interface{}, length int) (buf []byte, err error) {
switch value := data.(type) {
case encoding.BinaryMarshaler:
buf, err = value.MarshalBinary()
case string:
buf = []byte(value)
case []byte:
buf = value
case unsafe.Pointer:
err = xerrors.New("can't marshal from unsafe.Pointer")
default:
var wr bytes.Buffer
err = binary.Write(&wr, internal.NativeEndian, value)
if err != nil {
err = xerrors.Errorf("encoding %T: %v", value, err)
}
buf = wr.Bytes()
}
if err != nil {
return nil, err
}
if len(buf) != length {
return nil, xerrors.Errorf("%T doesn't marshal to %d bytes", data, length)
}
return buf, nil
}
func makeBuffer(dst interface{}, length int) (internal.Pointer, []byte) {
if ptr, ok := dst.(unsafe.Pointer); ok {
return internal.NewPointer(ptr), nil
}
buf := make([]byte, length)
return internal.NewSlicePointer(buf), buf
}
func unmarshalBytes(data interface{}, buf []byte) error {
switch value := data.(type) {
case unsafe.Pointer:
sh := &reflect.SliceHeader{
Data: uintptr(value),
Len: len(buf),
Cap: len(buf),
}
dst := *(*[]byte)(unsafe.Pointer(sh))
copy(dst, buf)
runtime.KeepAlive(value)
return nil
case encoding.BinaryUnmarshaler:
return value.UnmarshalBinary(buf)
case *string:
*value = string(buf)
return nil
case *[]byte:
*value = buf
return nil
case string:
return xerrors.New("require pointer to string")
case []byte:
return xerrors.New("require pointer to []byte")
default:
rd := bytes.NewReader(buf)
if err := binary.Read(rd, internal.NativeEndian, value); err != nil {
return xerrors.Errorf("decoding %T: %v", value, err)
}
return nil
}
}
// marshalPerCPUValue encodes a slice containing one value per
// possible CPU into a buffer of bytes.
//
// Values are initialized to zero if the slice has less elements than CPUs.
//
// slice must have a type like []elementType.
func marshalPerCPUValue(slice interface{}, elemLength int) (internal.Pointer, error) {
sliceType := reflect.TypeOf(slice)
if sliceType.Kind() != reflect.Slice {
return internal.Pointer{}, xerrors.New("per-CPU value requires slice")
}
possibleCPUs, err := internal.PossibleCPUs()
if err != nil {
return internal.Pointer{}, err
}
sliceValue := reflect.ValueOf(slice)
sliceLen := sliceValue.Len()
if sliceLen > possibleCPUs {
return internal.Pointer{}, xerrors.Errorf("per-CPU value exceeds number of CPUs")
}
alignedElemLength := align(elemLength, 8)
buf := make([]byte, alignedElemLength*possibleCPUs)
for i := 0; i < sliceLen; i++ {
elem := sliceValue.Index(i).Interface()
elemBytes, err := marshalBytes(elem, elemLength)
if err != nil {
return internal.Pointer{}, err
}
offset := i * alignedElemLength
copy(buf[offset:offset+elemLength], elemBytes)
}
return internal.NewSlicePointer(buf), nil
}
// unmarshalPerCPUValue decodes a buffer into a slice containing one value per
// possible CPU.
//
// valueOut must have a type like *[]elementType
func unmarshalPerCPUValue(slicePtr interface{}, elemLength int, buf []byte) error {
slicePtrType := reflect.TypeOf(slicePtr)
if slicePtrType.Kind() != reflect.Ptr || slicePtrType.Elem().Kind() != reflect.Slice {
return xerrors.Errorf("per-cpu value requires pointer to slice")
}
possibleCPUs, err := internal.PossibleCPUs()
if err != nil {
return err
}
sliceType := slicePtrType.Elem()
slice := reflect.MakeSlice(sliceType, possibleCPUs, possibleCPUs)
sliceElemType := sliceType.Elem()
sliceElemIsPointer := sliceElemType.Kind() == reflect.Ptr
if sliceElemIsPointer {
sliceElemType = sliceElemType.Elem()
}
step := len(buf) / possibleCPUs
if step < elemLength {
return xerrors.Errorf("per-cpu element length is larger than available data")
}
for i := 0; i < possibleCPUs; i++ {
var elem interface{}
if sliceElemIsPointer {
newElem := reflect.New(sliceElemType)
slice.Index(i).Set(newElem)
elem = newElem.Interface()
} else {
elem = slice.Index(i).Addr().Interface()
}
// Make a copy, since unmarshal can hold on to itemBytes
elemBytes := make([]byte, elemLength)
copy(elemBytes, buf[:elemLength])
err := unmarshalBytes(elem, elemBytes)
if err != nil {
return xerrors.Errorf("cpu %d: %w", i, err)
}
buf = buf[step:]
}
reflect.ValueOf(slicePtr).Elem().Set(slice)
return nil
}
func align(n, alignment int) int {
return (int(n) + alignment - 1) / alignment * alignment
}
eBPF
-------
[![](https://godoc.org/github.com/cilium/ebpf?status.svg)](https://godoc.org/github.com/cilium/ebpf)
eBPF is a pure Go library that provides utilities for loading, compiling, and debugging eBPF programs. It has minimal external dependencies and is intended to be used in long running processes.
[ebpf/asm](https://godoc.org/github.com/cilium/ebpf/asm) contains a basic assembler.
The library is maintained by [Cloudflare](https://www.cloudflare.com) and [Cilium](https://www.cilium.io). Feel free to [join](https://cilium.herokuapp.com/) the [libbpf-go](https://cilium.slack.com/messages/libbpf-go) channel on Slack.
## Current status
The package is production ready, but **the API is explicitly unstable
right now**. Expect to update your code if you want to follow along.
## Requirements
* A version of Go that is [supported by upstream](https://golang.org/doc/devel/release.html#policy)
* Linux 4.9, 4.19 or 5.4 (versions in-between should work, but are not tested)
## Useful resources
* [Cilium eBPF documentation](https://cilium.readthedocs.io/en/latest/bpf/#bpf-guide) (recommended)
* [Linux documentation on BPF](http://elixir.free-electrons.com/linux/latest/source/Documentation/networking/filter.txt)
* [eBPF features by Linux version](https://github.com/iovisor/bcc/blob/master/docs/kernel-versions.md)
#!/bin/bash
# Test the current package under a different kernel.
# Requires virtme and qemu to be installed.
set -eu
set -o pipefail
if [[ "${1:-}" = "--in-vm" ]]; then
shift
mount -t bpf bpf /sys/fs/bpf
export CGO_ENABLED=0
export GOFLAGS=-mod=readonly
export GOPATH=/run/go-path
export GOPROXY=file:///run/go-root/pkg/mod/cache/download
export GOCACHE=/run/go-cache
echo Running tests...
/usr/local/bin/go test -coverprofile="$1/coverage.txt" -covermode=atomic -v ./...
touch "$1/success"
exit 0
fi
# Pull all dependencies, so that we can run tests without the
# vm having network access.
go mod download
# Use sudo if /dev/kvm isn't accessible by the current user.
sudo=""
if [[ ! -r /dev/kvm || ! -w /dev/kvm ]]; then
sudo="sudo"
fi
readonly sudo
readonly kernel_version="${1:-}"
if [[ -z "${kernel_version}" ]]; then
echo "Expecting kernel version as first argument"
exit 1
fi
readonly kernel="linux-${kernel_version}.bz"
readonly output="$(mktemp -d)"
readonly tmp_dir="${TMPDIR:-$(mktemp -d)}"
test -e "${tmp_dir}/${kernel}" || {
echo Fetching "${kernel}"
curl --fail -L "https://github.com/cilium/ci-kernels/blob/master/${kernel}?raw=true" -o "${tmp_dir}/${kernel}"
}
echo Testing on "${kernel_version}"
$sudo virtme-run --kimg "${tmp_dir}/${kernel}" --memory 512M --pwd \
--rwdir=/run/output="${output}" \
--rodir=/run/go-path="$(go env GOPATH)" \
--rwdir=/run/go-cache="$(go env GOCACHE)" \
--script-sh "$(realpath "$0") --in-vm /run/output"
if [[ ! -e "${output}/success" ]]; then
echo "Test failed on ${kernel_version}"
exit 1
else
echo "Test successful on ${kernel_version}"
if [[ -v CODECOV_TOKEN ]]; then
curl --fail -s https://codecov.io/bash > "${tmp_dir}/codecov.sh"
chmod +x "${tmp_dir}/codecov.sh"
"${tmp_dir}/codecov.sh" -f "${output}/coverage.txt"
fi
fi
$sudo rm -r "${output}"
package ebpf
//go:generate stringer -output types_string.go -type=MapType,ProgramType
// MapType indicates the type map structure
// that will be initialized in the kernel.
type MapType uint32
// All the various map types that can be created
const (
UnspecifiedMap MapType = iota
// Hash is a hash map
Hash
// Array is an array map
Array
// ProgramArray - A program array map is a special kind of array map whose map
// values contain only file descriptors referring to other eBPF
// programs. Thus, both the key_size and value_size must be
// exactly four bytes. This map is used in conjunction with the
// TailCall helper.
ProgramArray
// PerfEventArray - A perf event array is used in conjunction with PerfEventRead
// and PerfEventOutput calls, to read the raw bpf_perf_data from the registers.
PerfEventArray
// PerCPUHash - This data structure is useful for people who have high performance
// network needs and can reconcile adds at the end of some cycle, so that
// hashes can be lock free without the use of XAdd, which can be costly.
PerCPUHash
// PerCPUArray - This data structure is useful for people who have high performance
// network needs and can reconcile adds at the end of some cycle, so that
// hashes can be lock free without the use of XAdd, which can be costly.
// Each CPU gets a copy of this hash, the contents of all of which can be reconciled
// later.
PerCPUArray
// StackTrace - This holds whole user and kernel stack traces, it can be retrieved with
// GetStackID
StackTrace
// CGroupArray - This is a very niche structure used to help SKBInCGroup determine
// if an skb is from a socket belonging to a specific cgroup
CGroupArray
// LRUHash - This allows you to create a small hash structure that will purge the
// least recently used items rather than thow an error when you run out of memory
LRUHash
// LRUCPUHash - This is NOT like PerCPUHash, this structure is shared among the CPUs,
// it has more to do with including the CPU id with the LRU calculation so that if a
// particular CPU is using a value over-and-over again, then it will be saved, but if
// a value is being retrieved a lot but sparsely across CPUs it is not as important, basically
// giving weight to CPU locality over overall usage.
LRUCPUHash
// LPMTrie - This is an implementation of Longest-Prefix-Match Trie structure. It is useful,
// for storing things like IP addresses which can be bit masked allowing for keys of differing
// values to refer to the same reference based on their masks. See wikipedia for more details.
LPMTrie
// ArrayOfMaps - Each item in the array is another map. The inner map mustn't be a map of maps
// itself.
ArrayOfMaps
// HashOfMaps - Each item in the hash map is another map. The inner map mustn't be a map of maps
// itself.
HashOfMaps
// DevMap - Specialized map to store references to network devices.
DevMap
// SockMap - Specialized map to store references to sockets.
SockMap
// CPUMap - Specialized map to store references to CPUs.
CPUMap
// XSKMap - Specialized map for XDP programs to store references to open sockets.
XSKMap
// SockHash - Specialized hash to store references to sockets.
SockHash
// CGroupStorage - Special map for CGroups.
CGroupStorage
// ReusePortSockArray - Specialized map to store references to sockets that can be reused.
ReusePortSockArray
// PerCPUCGroupStorage - Special per CPU map for CGroups.
PerCPUCGroupStorage
// Queue - FIFO storage for BPF programs.
Queue
// Stack - LIFO storage for BPF programs.
Stack
// SkStorage - Specialized map for local storage at SK for BPF programs.
SkStorage
// DevMapHash - Hash-based indexing scheme for references to network devices.
DevMapHash
)
// hasPerCPUValue returns true if the Map stores a value per CPU.
func (mt MapType) hasPerCPUValue() bool {
if mt == PerCPUHash || mt == PerCPUArray {
return true
}
return false
}
const (
_MapCreate = iota
_MapLookupElem
_MapUpdateElem
_MapDeleteElem
_MapGetNextKey
_ProgLoad
_ObjPin
_ObjGet
_ProgAttach
_ProgDetach
_ProgTestRun
_ProgGetNextID
_MapGetNextID
_ProgGetFDByID
_MapGetFDByID
_ObjGetInfoByFD
_ProgQuery
_RawTracepointOpen
_BTFLoad
_BTFGetFDByID
_TaskFDQuery
_MapLookupAndDeleteElem
_MapFreeze
_BTFGetNextID
)
// ProgramType of the eBPF program
type ProgramType uint32
// eBPF program types
const (
// Unrecognized program type
UnspecifiedProgram ProgramType = iota
// SocketFilter socket or seccomp filter
SocketFilter
// Kprobe program
Kprobe
// SchedCLS traffic control shaper
SchedCLS
// SchedACT routing control shaper
SchedACT
// TracePoint program
TracePoint
// XDP program
XDP
// PerfEvent program
PerfEvent
// CGroupSKB program
CGroupSKB
// CGroupSock program
CGroupSock
// LWTIn program
LWTIn
// LWTOut program
LWTOut
// LWTXmit program
LWTXmit
// SockOps program
SockOps
// SkSKB program
SkSKB
// CGroupDevice program
CGroupDevice
// SkMsg program
SkMsg
// RawTracepoint program
RawTracepoint
// CGroupSockAddr program
CGroupSockAddr
// LWTSeg6Local program
LWTSeg6Local
// LircMode2 program
LircMode2
// SkReuseport program
SkReuseport
// FlowDissector program
FlowDissector
// CGroupSysctl program
CGroupSysctl
// RawTracepointWritable program
RawTracepointWritable
// CGroupSockopt program
CGroupSockopt
// Tracing program
Tracing
)
// AttachType of the eBPF program, needed to differentiate allowed context accesses in
// some newer program types like CGroupSockAddr. Should be set to AttachNone if not required.
// Will cause invalid argument (EINVAL) at program load time if set incorrectly.
type AttachType uint32
// AttachNone is an alias for AttachCGroupInetIngress for readability reasons
const AttachNone AttachType = 0
const (
AttachCGroupInetIngress AttachType = iota
AttachCGroupInetEgress
AttachCGroupInetSockCreate
AttachCGroupSockOps
AttachSkSKBStreamParser
AttachSkSKBStreamVerdict
AttachCGroupDevice
AttachSkMsgVerdict
AttachCGroupInet4Bind
AttachCGroupInet6Bind
AttachCGroupInet4Connect
AttachCGroupInet6Connect
AttachCGroupInet4PostBind
AttachCGroupInet6PostBind
AttachCGroupUDP4Sendmsg
AttachCGroupUDP6Sendmsg
AttachLircMode2
AttachFlowDissector
AttachCGroupSysctl
AttachCGroupUDP4Recvmsg
AttachCGroupUDP6Recvmsg
AttachCGroupGetsockopt
AttachCGroupSetsockopt
AttachTraceRawTp
AttachTraceFEntry
AttachTraceFExit
)
// AttachFlags of the eBPF program used in BPF_PROG_ATTACH command
type AttachFlags uint32
// Code generated by "stringer -output types_string.go -type=MapType,ProgramType"; DO NOT EDIT.
package ebpf
import "strconv"
func _() {
// An "invalid array index" compiler error signifies that the constant values have changed.
// Re-run the stringer command to generate them again.
var x [1]struct{}
_ = x[UnspecifiedMap-0]
_ = x[Hash-1]
_ = x[Array-2]
_ = x[ProgramArray-3]
_ = x[PerfEventArray-4]
_ = x[PerCPUHash-5]
_ = x[PerCPUArray-6]
_ = x[StackTrace-7]
_ = x[CGroupArray-8]
_ = x[LRUHash-9]
_ = x[LRUCPUHash-10]
_ = x[LPMTrie-11]
_ = x[ArrayOfMaps-12]
_ = x[HashOfMaps-13]
_ = x[DevMap-14]
_ = x[SockMap-15]
_ = x[CPUMap-16]
_ = x[XSKMap-17]
_ = x[SockHash-18]
_ = x[CGroupStorage-19]
_ = x[ReusePortSockArray-20]
_ = x[PerCPUCGroupStorage-21]
_ = x[Queue-22]
_ = x[Stack-23]
_ = x[SkStorage-24]
_ = x[DevMapHash-25]
}
const _MapType_name = "UnspecifiedMapHashArrayProgramArrayPerfEventArrayPerCPUHashPerCPUArrayStackTraceCGroupArrayLRUHashLRUCPUHashLPMTrieArrayOfMapsHashOfMapsDevMapSockMapCPUMapXSKMapSockHashCGroupStorageReusePortSockArrayPerCPUCGroupStorageQueueStackSkStorageDevMapHash"
var _MapType_index = [...]uint8{0, 14, 18, 23, 35, 49, 59, 70, 80, 91, 98, 108, 115, 126, 136, 142, 149, 155, 161, 169, 182, 200, 219, 224, 229, 238, 248}
func (i MapType) String() string {
if i >= MapType(len(_MapType_index)-1) {
return "MapType(" + strconv.FormatInt(int64(i), 10) + ")"
}
return _MapType_name[_MapType_index[i]:_MapType_index[i+1]]
}
func _() {
// An "invalid array index" compiler error signifies that the constant values have changed.
// Re-run the stringer command to generate them again.
var x [1]struct{}
_ = x[UnspecifiedProgram-0]
_ = x[SocketFilter-1]
_ = x[Kprobe-2]
_ = x[SchedCLS-3]
_ = x[SchedACT-4]
_ = x[TracePoint-5]
_ = x[XDP-6]
_ = x[PerfEvent-7]
_ = x[CGroupSKB-8]
_ = x[CGroupSock-9]
_ = x[LWTIn-10]
_ = x[LWTOut-11]
_ = x[LWTXmit-12]
_ = x[SockOps-13]
_ = x[SkSKB-14]
_ = x[CGroupDevice-15]
_ = x[SkMsg-16]
_ = x[RawTracepoint-17]
_ = x[CGroupSockAddr-18]
_ = x[LWTSeg6Local-19]
_ = x[LircMode2-20]
_ = x[SkReuseport-21]
_ = x[FlowDissector-22]
_ = x[CGroupSysctl-23]
_ = x[RawTracepointWritable-24]
_ = x[CGroupSockopt-25]
_ = x[Tracing-26]
}
const _ProgramType_name = "UnspecifiedProgramSocketFilterKprobeSchedCLSSchedACTTracePointXDPPerfEventCGroupSKBCGroupSockLWTInLWTOutLWTXmitSockOpsSkSKBCGroupDeviceSkMsgRawTracepointCGroupSockAddrLWTSeg6LocalLircMode2SkReuseportFlowDissectorCGroupSysctlRawTracepointWritableCGroupSockoptTracing"
var _ProgramType_index = [...]uint16{0, 18, 30, 36, 44, 52, 62, 65, 74, 83, 93, 98, 104, 111, 118, 123, 135, 140, 153, 167, 179, 188, 199, 212, 224, 245, 258, 265}
func (i ProgramType) String() string {
if i >= ProgramType(len(_ProgramType_index)-1) {
return "ProgramType(" + strconv.FormatInt(int64(i), 10) + ")"
}
return _ProgramType_name[_ProgramType_index[i]:_ProgramType_index[i+1]]
}
dist: xenial dist: bionic
language: go language: go
go: go:
- 1.11.x - 1.11.x
...@@ -13,6 +13,20 @@ matrix: ...@@ -13,6 +13,20 @@ matrix:
script: script:
- make BUILDTAGS="${BUILDTAGS}" all - make BUILDTAGS="${BUILDTAGS}" all
- sudo PATH="$PATH" make localintegration RUNC_USE_SYSTEMD=1 - sudo PATH="$PATH" make localintegration RUNC_USE_SYSTEMD=1
- go: 1.12.x
env:
- VIRTUALBOX_VERSION=6.0
- VAGRANT_VERSION=2.2.6
- FEDORA_VERSION=31
before_install:
- cat /proc/cpuinfo
- wget -q https://www.virtualbox.org/download/oracle_vbox_2016.asc -O- | sudo apt-key add - && sudo sh -c "echo deb https://download.virtualbox.org/virtualbox/debian $(lsb_release -cs) contrib >> /etc/apt/sources.list" && sudo apt-get update && sudo apt-get install -yq build-essential gcc make linux-headers-$(uname -r) virtualbox-${VIRTUALBOX_VERSION} && sudo usermod -aG vboxusers $(whoami)
- wget https://releases.hashicorp.com/vagrant/${VAGRANT_VERSION}/vagrant_${VAGRANT_VERSION}_$(uname -m).deb && sudo dpkg -i vagrant_${VAGRANT_VERSION}_$(uname -m).deb
- vagrant init bento/fedora-${FEDORA_VERSION} && vagrant up && mkdir -p ~/.ssh && vagrant ssh-config >> ~/.ssh/config
- ssh default sudo dnf install -y podman
script:
- ssh default sudo podman build -t test /vagrant
- ssh default sudo podman run --privileged --cgroupns=private test make localunittest
allow_failures: allow_failures:
- go: tip - go: tip
......
...@@ -3,6 +3,7 @@ ...@@ -3,6 +3,7 @@
test unittest integration \ test unittest integration \
cross localcross cross localcross
CONTAINER_ENGINE := docker
GO := go GO := go
SOURCES := $(shell find . 2>&1 | grep -E '.*\.(c|h|go)$$') SOURCES := $(shell find . 2>&1 | grep -E '.*\.(c|h|go)$$')
...@@ -40,14 +41,14 @@ contrib/cmd/recvtty/recvtty: $(SOURCES) ...@@ -40,14 +41,14 @@ contrib/cmd/recvtty/recvtty: $(SOURCES)
$(GO) build -buildmode=pie $(EXTRA_FLAGS) -ldflags "-X main.gitCommit=${COMMIT} -X main.version=${VERSION} $(EXTRA_LDFLAGS)" -tags "$(BUILDTAGS)" -o contrib/cmd/recvtty/recvtty ./contrib/cmd/recvtty $(GO) build -buildmode=pie $(EXTRA_FLAGS) -ldflags "-X main.gitCommit=${COMMIT} -X main.version=${VERSION} $(EXTRA_LDFLAGS)" -tags "$(BUILDTAGS)" -o contrib/cmd/recvtty/recvtty ./contrib/cmd/recvtty
static: $(SOURCES) static: $(SOURCES)
CGO_ENABLED=1 $(GO) build $(EXTRA_FLAGS) -tags "$(BUILDTAGS) netgo osusergo static_build" -installsuffix netgo -ldflags "-w -extldflags -static -X main.gitCommit=${COMMIT} -X main.version=${VERSION} $(EXTRA_LDFLAGS)" -o runc . CGO_ENABLED=1 $(GO) build $(EXTRA_FLAGS) -tags "$(BUILDTAGS) netgo osusergo" -installsuffix netgo -ldflags "-w -extldflags -static -X main.gitCommit=${COMMIT} -X main.version=${VERSION} $(EXTRA_LDFLAGS)" -o runc .
CGO_ENABLED=1 $(GO) build $(EXTRA_FLAGS) -tags "$(BUILDTAGS) netgo osusergo static_build" -installsuffix netgo -ldflags "-w -extldflags -static -X main.gitCommit=${COMMIT} -X main.version=${VERSION} $(EXTRA_LDFLAGS)" -o contrib/cmd/recvtty/recvtty ./contrib/cmd/recvtty CGO_ENABLED=1 $(GO) build $(EXTRA_FLAGS) -tags "$(BUILDTAGS) netgo osusergo" -installsuffix netgo -ldflags "-w -extldflags -static -X main.gitCommit=${COMMIT} -X main.version=${VERSION} $(EXTRA_LDFLAGS)" -o contrib/cmd/recvtty/recvtty ./contrib/cmd/recvtty
release: release:
script/release.sh -r release/$(VERSION) -v $(VERSION) script/release.sh -r release/$(VERSION) -v $(VERSION)
dbuild: runcimage dbuild: runcimage
docker run ${DOCKER_RUN_PROXY} --rm -v $(CURDIR):/go/src/$(PROJECT) --privileged $(RUNC_IMAGE) make clean all $(CONTAINER_ENGINE) run ${CONTAINER_ENGINE_RUN_FLAGS} --rm -v $(CURDIR):/go/src/$(PROJECT) --privileged $(RUNC_IMAGE) make clean all
lint: lint:
$(GO) vet $(allpackages) $(GO) vet $(allpackages)
...@@ -57,7 +58,7 @@ man: ...@@ -57,7 +58,7 @@ man:
man/md2man-all.sh man/md2man-all.sh
runcimage: runcimage:
docker build ${DOCKER_BUILD_PROXY} -t $(RUNC_IMAGE) . $(CONTAINER_ENGINE) build ${CONTAINER_ENGINE_BUILD_FLAGS} -t $(RUNC_IMAGE) .
test: test:
make unittest integration rootlessintegration make unittest integration rootlessintegration
...@@ -66,25 +67,25 @@ localtest: ...@@ -66,25 +67,25 @@ localtest:
make localunittest localintegration localrootlessintegration make localunittest localintegration localrootlessintegration
unittest: runcimage unittest: runcimage
docker run ${DOCKER_RUN_PROXY} -t --privileged --rm -v /lib/modules:/lib/modules:ro -v $(CURDIR):/go/src/$(PROJECT) $(RUNC_IMAGE) make localunittest TESTFLAGS=${TESTFLAGS} $(CONTAINER_ENGINE) run ${CONTAINER_ENGINE_RUN_FLAGS} -t --privileged --rm -v /lib/modules:/lib/modules:ro -v $(CURDIR):/go/src/$(PROJECT) $(RUNC_IMAGE) make localunittest TESTFLAGS=${TESTFLAGS}
localunittest: all localunittest: all
$(GO) test -timeout 3m -tags "$(BUILDTAGS)" ${TESTFLAGS} -v $(allpackages) $(GO) test -timeout 3m -tags "$(BUILDTAGS)" ${TESTFLAGS} -v $(allpackages)
integration: runcimage integration: runcimage
docker run ${DOCKER_RUN_PROXY} -t --privileged --rm -v /lib/modules:/lib/modules:ro -v $(CURDIR):/go/src/$(PROJECT) $(RUNC_IMAGE) make localintegration TESTPATH=${TESTPATH} $(CONTAINER_ENGINE) run ${CONTAINER_ENGINE_RUN_FLAGS} -t --privileged --rm -v /lib/modules:/lib/modules:ro -v $(CURDIR):/go/src/$(PROJECT) $(RUNC_IMAGE) make localintegration TESTPATH=${TESTPATH}
localintegration: all localintegration: all
bats -t tests/integration${TESTPATH} bats -t tests/integration${TESTPATH}
rootlessintegration: runcimage rootlessintegration: runcimage
docker run ${DOCKER_RUN_PROXY} -t --privileged --rm -v $(CURDIR):/go/src/$(PROJECT) $(RUNC_IMAGE) make localrootlessintegration $(CONTAINER_ENGINE) run ${CONTAINER_ENGINE_RUN_FLAGS} -t --privileged --rm -v $(CURDIR):/go/src/$(PROJECT) $(RUNC_IMAGE) make localrootlessintegration
localrootlessintegration: all localrootlessintegration: all
tests/rootless.sh tests/rootless.sh
shell: runcimage shell: runcimage
docker run ${DOCKER_RUN_PROXY} -ti --privileged --rm -v $(CURDIR):/go/src/$(PROJECT) $(RUNC_IMAGE) bash $(CONTAINER_ENGINE) run ${CONTAINER_ENGINE_RUN_FLAGS} -ti --privileged --rm -v $(CURDIR):/go/src/$(PROJECT) $(RUNC_IMAGE) bash
install: install:
install -D -m0755 runc $(BINDIR)/runc install -D -m0755 runc $(BINDIR)/runc
...@@ -119,7 +120,7 @@ validate: ...@@ -119,7 +120,7 @@ validate:
ci: validate test release ci: validate test release
cross: runcimage cross: runcimage
docker run ${DOCKER_RUN_PROXY} -e BUILDTAGS="$(BUILDTAGS)" --rm -v $(CURDIR):/go/src/$(PROJECT) $(RUNC_IMAGE) make localcross $(CONTAINER_ENGINE) run ${CONTAINER_ENGINE_RUN_FLAGS} -e BUILDTAGS="$(BUILDTAGS)" --rm -v $(CURDIR):/go/src/$(PROJECT) $(RUNC_IMAGE) make localcross
localcross: localcross:
CGO_ENABLED=1 GOARCH=arm GOARM=6 CC=arm-linux-gnueabi-gcc $(GO) build -buildmode=pie $(EXTRA_FLAGS) -ldflags "-X main.gitCommit=${COMMIT} -X main.version=${VERSION} $(EXTRA_LDFLAGS)" -tags "$(BUILDTAGS)" -o runc-armel . CGO_ENABLED=1 GOARCH=arm GOARM=6 CC=arm-linux-gnueabi-gcc $(GO) build -buildmode=pie $(EXTRA_FLAGS) -ldflags "-X main.gitCommit=${COMMIT} -X main.version=${VERSION} $(EXTRA_LDFLAGS)" -tags "$(BUILDTAGS)" -o runc-armel .
......
...@@ -16,9 +16,13 @@ This means that `runc` 1.0.0 should implement the 1.0 version of the specificati ...@@ -16,9 +16,13 @@ This means that `runc` 1.0.0 should implement the 1.0 version of the specificati
You can find official releases of `runc` on the [release](https://github.com/opencontainers/runc/releases) page. You can find official releases of `runc` on the [release](https://github.com/opencontainers/runc/releases) page.
Currently, the following features are not considered to be production-ready:
* Support for cgroup v2
## Security ## Security
Reporting process and disclosure communications are outlined in [/org/security](https://github.com/opencontainers/org/blob/master/security/) The reporting process and disclosure communications are outlined in [/org/security](https://github.com/opencontainers/org/blob/master/security/).
## Building ## Building
...@@ -229,7 +233,14 @@ runc delete mycontainerid ...@@ -229,7 +233,14 @@ runc delete mycontainerid
This allows higher level systems to augment the containers creation logic with setup of various settings after the container is created and/or before it is deleted. For example, the container's network stack is commonly set up after `create` but before `start`. This allows higher level systems to augment the containers creation logic with setup of various settings after the container is created and/or before it is deleted. For example, the container's network stack is commonly set up after `create` but before `start`.
#### Rootless containers #### Rootless containers
`runc` has the ability to run containers without root privileges. This is called `rootless`. You need to pass some parameters to `runc` in order to run rootless containers. See below and compare with the previous version. Run the following commands as an ordinary user: `runc` has the ability to run containers without root privileges. This is called `rootless`. You need to pass some parameters to `runc` in order to run rootless containers. See below and compare with the previous version.
**Note:** In order to use this feature, "User Namespaces" must be compiled and enabled in your kernel. There are various ways to do this depending on your distribution:
- Confirm `CONFIG_USER_NS=y` is set in your kernel configuration (normally found in `/proc/config.gz`)
- Arch/Debian: `echo 1 > /proc/sys/kernel/unprivileged_userns_clone`
- RHEL/CentOS 7: `echo 28633 > /proc/sys/user/max_user_namespaces`
Run the following commands as an ordinary user:
```bash ```bash
# Same as the first example # Same as the first example
mkdir ~/mycontainer mkdir ~/mycontainer
......
# Security
The reporting process and disclosure communications are outlined in [/org/security](https://github.com/opencontainers/org/blob/master/security/).
...@@ -12,125 +12,12 @@ import ( ...@@ -12,125 +12,12 @@ import (
"github.com/opencontainers/runc/libcontainer" "github.com/opencontainers/runc/libcontainer"
"github.com/opencontainers/runc/libcontainer/cgroups" "github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/intelrdt" "github.com/opencontainers/runc/libcontainer/intelrdt"
"github.com/opencontainers/runc/types"
"github.com/sirupsen/logrus" "github.com/sirupsen/logrus"
"github.com/urfave/cli" "github.com/urfave/cli"
) )
// event struct for encoding the event data to json.
type event struct {
Type string `json:"type"`
ID string `json:"id"`
Data interface{} `json:"data,omitempty"`
}
// stats is the runc specific stats structure for stability when encoding and decoding stats.
type stats struct {
CPU cpu `json:"cpu"`
Memory memory `json:"memory"`
Pids pids `json:"pids"`
Blkio blkio `json:"blkio"`
Hugetlb map[string]hugetlb `json:"hugetlb"`
IntelRdt intelRdt `json:"intel_rdt"`
}
type hugetlb struct {
Usage uint64 `json:"usage,omitempty"`
Max uint64 `json:"max,omitempty"`
Failcnt uint64 `json:"failcnt"`
}
type blkioEntry struct {
Major uint64 `json:"major,omitempty"`
Minor uint64 `json:"minor,omitempty"`
Op string `json:"op,omitempty"`
Value uint64 `json:"value,omitempty"`
}
type blkio struct {
IoServiceBytesRecursive []blkioEntry `json:"ioServiceBytesRecursive,omitempty"`
IoServicedRecursive []blkioEntry `json:"ioServicedRecursive,omitempty"`
IoQueuedRecursive []blkioEntry `json:"ioQueueRecursive,omitempty"`
IoServiceTimeRecursive []blkioEntry `json:"ioServiceTimeRecursive,omitempty"`
IoWaitTimeRecursive []blkioEntry `json:"ioWaitTimeRecursive,omitempty"`
IoMergedRecursive []blkioEntry `json:"ioMergedRecursive,omitempty"`
IoTimeRecursive []blkioEntry `json:"ioTimeRecursive,omitempty"`
SectorsRecursive []blkioEntry `json:"sectorsRecursive,omitempty"`
}
type pids struct {
Current uint64 `json:"current,omitempty"`
Limit uint64 `json:"limit,omitempty"`
}
type throttling struct {
Periods uint64 `json:"periods,omitempty"`
ThrottledPeriods uint64 `json:"throttledPeriods,omitempty"`
ThrottledTime uint64 `json:"throttledTime,omitempty"`
}
type cpuUsage struct {
// Units: nanoseconds.
Total uint64 `json:"total,omitempty"`
Percpu []uint64 `json:"percpu,omitempty"`
Kernel uint64 `json:"kernel"`
User uint64 `json:"user"`
}
type cpu struct {
Usage cpuUsage `json:"usage,omitempty"`
Throttling throttling `json:"throttling,omitempty"`
}
type memoryEntry struct {
Limit uint64 `json:"limit"`
Usage uint64 `json:"usage,omitempty"`
Max uint64 `json:"max,omitempty"`
Failcnt uint64 `json:"failcnt"`
}
type memory struct {
Cache uint64 `json:"cache,omitempty"`
Usage memoryEntry `json:"usage,omitempty"`
Swap memoryEntry `json:"swap,omitempty"`
Kernel memoryEntry `json:"kernel,omitempty"`
KernelTCP memoryEntry `json:"kernelTCP,omitempty"`
Raw map[string]uint64 `json:"raw,omitempty"`
}
type l3CacheInfo struct {
CbmMask string `json:"cbm_mask,omitempty"`
MinCbmBits uint64 `json:"min_cbm_bits,omitempty"`
NumClosids uint64 `json:"num_closids,omitempty"`
}
type memBwInfo struct {
BandwidthGran uint64 `json:"bandwidth_gran,omitempty"`
DelayLinear uint64 `json:"delay_linear,omitempty"`
MinBandwidth uint64 `json:"min_bandwidth,omitempty"`
NumClosids uint64 `json:"num_closids,omitempty"`
}
type intelRdt struct {
// The read-only L3 cache information
L3CacheInfo *l3CacheInfo `json:"l3_cache_info,omitempty"`
// The read-only L3 cache schema in root
L3CacheSchemaRoot string `json:"l3_cache_schema_root,omitempty"`
// The L3 cache schema in 'container_id' group
L3CacheSchema string `json:"l3_cache_schema,omitempty"`
// The read-only memory bandwidth information
MemBwInfo *memBwInfo `json:"mem_bw_info,omitempty"`
// The read-only memory bandwidth schema in root
MemBwSchemaRoot string `json:"mem_bw_schema_root,omitempty"`
// The memory bandwidth schema in 'container_id' group
MemBwSchema string `json:"mem_bw_schema,omitempty"`
}
var eventsCommand = cli.Command{ var eventsCommand = cli.Command{
Name: "events", Name: "events",
Usage: "display container events such as OOM notifications, cpu, memory, and IO usage statistics", Usage: "display container events such as OOM notifications, cpu, memory, and IO usage statistics",
...@@ -164,7 +51,7 @@ information is displayed once every 5 seconds.`, ...@@ -164,7 +51,7 @@ information is displayed once every 5 seconds.`,
} }
var ( var (
stats = make(chan *libcontainer.Stats, 1) stats = make(chan *libcontainer.Stats, 1)
events = make(chan *event, 1024) events = make(chan *types.Event, 1024)
group = &sync.WaitGroup{} group = &sync.WaitGroup{}
) )
group.Add(1) group.Add(1)
...@@ -182,7 +69,7 @@ information is displayed once every 5 seconds.`, ...@@ -182,7 +69,7 @@ information is displayed once every 5 seconds.`,
if err != nil { if err != nil {
return err return err
} }
events <- &event{Type: "stats", ID: container.ID(), Data: convertLibcontainerStats(s)} events <- &types.Event{Type: "stats", ID: container.ID(), Data: convertLibcontainerStats(s)}
close(events) close(events)
group.Wait() group.Wait()
return nil return nil
...@@ -208,12 +95,12 @@ information is displayed once every 5 seconds.`, ...@@ -208,12 +95,12 @@ information is displayed once every 5 seconds.`,
// this means an oom event was received, if it is !ok then // this means an oom event was received, if it is !ok then
// the channel was closed because the container stopped and // the channel was closed because the container stopped and
// the cgroups no longer exist. // the cgroups no longer exist.
events <- &event{Type: "oom", ID: container.ID()} events <- &types.Event{Type: "oom", ID: container.ID()}
} else { } else {
n = nil n = nil
} }
case s := <-stats: case s := <-stats:
events <- &event{Type: "stats", ID: container.ID(), Data: convertLibcontainerStats(s)} events <- &types.Event{Type: "stats", ID: container.ID(), Data: convertLibcontainerStats(s)}
} }
if n == nil { if n == nil {
close(events) close(events)
...@@ -225,12 +112,12 @@ information is displayed once every 5 seconds.`, ...@@ -225,12 +112,12 @@ information is displayed once every 5 seconds.`,
}, },
} }
func convertLibcontainerStats(ls *libcontainer.Stats) *stats { func convertLibcontainerStats(ls *libcontainer.Stats) *types.Stats {
cg := ls.CgroupStats cg := ls.CgroupStats
if cg == nil { if cg == nil {
return nil return nil
} }
var s stats var s types.Stats
s.Pids.Current = cg.PidsStats.Current s.Pids.Current = cg.PidsStats.Current
s.Pids.Limit = cg.PidsStats.Limit s.Pids.Limit = cg.PidsStats.Limit
...@@ -258,7 +145,7 @@ func convertLibcontainerStats(ls *libcontainer.Stats) *stats { ...@@ -258,7 +145,7 @@ func convertLibcontainerStats(ls *libcontainer.Stats) *stats {
s.Blkio.IoTimeRecursive = convertBlkioEntry(cg.BlkioStats.IoTimeRecursive) s.Blkio.IoTimeRecursive = convertBlkioEntry(cg.BlkioStats.IoTimeRecursive)
s.Blkio.SectorsRecursive = convertBlkioEntry(cg.BlkioStats.SectorsRecursive) s.Blkio.SectorsRecursive = convertBlkioEntry(cg.BlkioStats.SectorsRecursive)
s.Hugetlb = make(map[string]hugetlb) s.Hugetlb = make(map[string]types.Hugetlb)
for k, v := range cg.HugetlbStats { for k, v := range cg.HugetlbStats {
s.Hugetlb[k] = convertHugtlb(v) s.Hugetlb[k] = convertHugtlb(v)
} }
...@@ -276,19 +163,20 @@ func convertLibcontainerStats(ls *libcontainer.Stats) *stats { ...@@ -276,19 +163,20 @@ func convertLibcontainerStats(ls *libcontainer.Stats) *stats {
} }
} }
s.NetworkInterfaces = ls.Interfaces
return &s return &s
} }
func convertHugtlb(c cgroups.HugetlbStats) hugetlb { func convertHugtlb(c cgroups.HugetlbStats) types.Hugetlb {
return hugetlb{ return types.Hugetlb{
Usage: c.Usage, Usage: c.Usage,
Max: c.MaxUsage, Max: c.MaxUsage,
Failcnt: c.Failcnt, Failcnt: c.Failcnt,
} }
} }
func convertMemoryEntry(c cgroups.MemoryData) memoryEntry { func convertMemoryEntry(c cgroups.MemoryData) types.MemoryEntry {
return memoryEntry{ return types.MemoryEntry{
Limit: c.Limit, Limit: c.Limit,
Usage: c.Usage, Usage: c.Usage,
Max: c.MaxUsage, Max: c.MaxUsage,
...@@ -296,10 +184,10 @@ func convertMemoryEntry(c cgroups.MemoryData) memoryEntry { ...@@ -296,10 +184,10 @@ func convertMemoryEntry(c cgroups.MemoryData) memoryEntry {
} }
} }
func convertBlkioEntry(c []cgroups.BlkioStatEntry) []blkioEntry { func convertBlkioEntry(c []cgroups.BlkioStatEntry) []types.BlkioEntry {
var out []blkioEntry var out []types.BlkioEntry
for _, e := range c { for _, e := range c {
out = append(out, blkioEntry{ out = append(out, types.BlkioEntry{
Major: e.Major, Major: e.Major,
Minor: e.Minor, Minor: e.Minor,
Op: e.Op, Op: e.Op,
...@@ -309,16 +197,16 @@ func convertBlkioEntry(c []cgroups.BlkioStatEntry) []blkioEntry { ...@@ -309,16 +197,16 @@ func convertBlkioEntry(c []cgroups.BlkioStatEntry) []blkioEntry {
return out return out
} }
func convertL3CacheInfo(i *intelrdt.L3CacheInfo) *l3CacheInfo { func convertL3CacheInfo(i *intelrdt.L3CacheInfo) *types.L3CacheInfo {
return &l3CacheInfo{ return &types.L3CacheInfo{
CbmMask: i.CbmMask, CbmMask: i.CbmMask,
MinCbmBits: i.MinCbmBits, MinCbmBits: i.MinCbmBits,
NumClosids: i.NumClosids, NumClosids: i.NumClosids,
} }
} }
func convertMemBwInfo(i *intelrdt.MemBwInfo) *memBwInfo { func convertMemBwInfo(i *intelrdt.MemBwInfo) *types.MemBwInfo {
return &memBwInfo{ return &types.MemBwInfo{
BandwidthGran: i.BandwidthGran, BandwidthGran: i.BandwidthGran,
DelayLinear: i.DelayLinear, DelayLinear: i.DelayLinear,
MinBandwidth: i.MinBandwidth, MinBandwidth: i.MinBandwidth,
......
...@@ -37,8 +37,18 @@ type Manager interface { ...@@ -37,8 +37,18 @@ type Manager interface {
// restore the object later. // restore the object later.
GetPaths() map[string]string GetPaths() map[string]string
// GetUnifiedPath returns the unified path when running in unified mode.
// The value corresponds to the all values of GetPaths() map.
//
// GetUnifiedPath returns error when running in hybrid mode as well as
// in legacy mode.
GetUnifiedPath() (string, error)
// Sets the cgroup as configured. // Sets the cgroup as configured.
Set(container *configs.Config) error Set(container *configs.Config) error
// Gets the cgroup as configured.
GetCgroups() (*configs.Cgroup, error)
} }
type NotFoundError struct { type NotFoundError struct {
......
// Package devicefilter containes eBPF device filter program
//
// The implementation is based on https://github.com/containers/crun/blob/0.10.2/src/libcrun/ebpf.c
//
// Although ebpf.c is originally licensed under LGPL-3.0-or-later, the author (Giuseppe Scrivano)
// agreed to relicense the file in Apache License 2.0: https://github.com/opencontainers/runc/issues/2144#issuecomment-543116397
package devicefilter
import (
"fmt"
"math"
"github.com/cilium/ebpf/asm"
"github.com/opencontainers/runc/libcontainer/configs"
"github.com/pkg/errors"
"golang.org/x/sys/unix"
)
const (
// license string format is same as kernel MODULE_LICENSE macro
license = "Apache"
)
// DeviceFilter returns eBPF device filter program and its license string
func DeviceFilter(devices []*configs.Device) (asm.Instructions, string, error) {
p := &program{}
p.init()
for i := len(devices) - 1; i >= 0; i-- {
if err := p.appendDevice(devices[i]); err != nil {
return nil, "", err
}
}
insts, err := p.finalize()
return insts, license, err
}
type program struct {
insts asm.Instructions
hasWildCard bool
blockID int
}
func (p *program) init() {
// struct bpf_cgroup_dev_ctx: https://elixir.bootlin.com/linux/v5.3.6/source/include/uapi/linux/bpf.h#L3423
/*
u32 access_type
u32 major
u32 minor
*/
// R2 <- type (lower 16 bit of u32 access_type at R1[0])
p.insts = append(p.insts,
asm.LoadMem(asm.R2, asm.R1, 0, asm.Half))
// R3 <- access (upper 16 bit of u32 access_type at R1[0])
p.insts = append(p.insts,
asm.LoadMem(asm.R3, asm.R1, 0, asm.Word),
// RSh: bitwise shift right
asm.RSh.Imm32(asm.R3, 16))
// R4 <- major (u32 major at R1[4])
p.insts = append(p.insts,
asm.LoadMem(asm.R4, asm.R1, 4, asm.Word))
// R5 <- minor (u32 minor at R1[8])
p.insts = append(p.insts,
asm.LoadMem(asm.R5, asm.R1, 8, asm.Word))
}
// appendDevice needs to be called from the last element of OCI linux.resources.devices to the head element.
func (p *program) appendDevice(dev *configs.Device) error {
if p.blockID < 0 {
return errors.New("the program is finalized")
}
if p.hasWildCard {
// All entries after wildcard entry are ignored
return nil
}
bpfType := int32(-1)
hasType := true
switch dev.Type {
case 'c':
bpfType = int32(unix.BPF_DEVCG_DEV_CHAR)
case 'b':
bpfType = int32(unix.BPF_DEVCG_DEV_BLOCK)
case 'a':
hasType = false
default:
// if not specified in OCI json, typ is set to DeviceTypeAll
return errors.Errorf("invalid DeviceType %q", string(dev.Type))
}
if dev.Major > math.MaxUint32 {
return errors.Errorf("invalid major %d", dev.Major)
}
if dev.Minor > math.MaxUint32 {
return errors.Errorf("invalid minor %d", dev.Major)
}
hasMajor := dev.Major >= 0 // if not specified in OCI json, major is set to -1
hasMinor := dev.Minor >= 0
bpfAccess := int32(0)
for _, r := range dev.Permissions {
switch r {
case 'r':
bpfAccess |= unix.BPF_DEVCG_ACC_READ
case 'w':
bpfAccess |= unix.BPF_DEVCG_ACC_WRITE
case 'm':
bpfAccess |= unix.BPF_DEVCG_ACC_MKNOD
default:
return errors.Errorf("unknown device access %v", r)
}
}
// If the access is rwm, skip the check.
hasAccess := bpfAccess != (unix.BPF_DEVCG_ACC_READ | unix.BPF_DEVCG_ACC_WRITE | unix.BPF_DEVCG_ACC_MKNOD)
blockSym := fmt.Sprintf("block-%d", p.blockID)
nextBlockSym := fmt.Sprintf("block-%d", p.blockID+1)
prevBlockLastIdx := len(p.insts) - 1
if hasType {
p.insts = append(p.insts,
// if (R2 != bpfType) goto next
asm.JNE.Imm(asm.R2, bpfType, nextBlockSym),
)
}
if hasAccess {
p.insts = append(p.insts,
// if (R3 & bpfAccess == 0 /* use R1 as a temp var */) goto next
asm.Mov.Reg32(asm.R1, asm.R3),
asm.And.Imm32(asm.R1, bpfAccess),
asm.JEq.Imm(asm.R1, 0, nextBlockSym),
)
}
if hasMajor {
p.insts = append(p.insts,
// if (R4 != major) goto next
asm.JNE.Imm(asm.R4, int32(dev.Major), nextBlockSym),
)
}
if hasMinor {
p.insts = append(p.insts,
// if (R5 != minor) goto next
asm.JNE.Imm(asm.R5, int32(dev.Minor), nextBlockSym),
)
}
if !hasType && !hasAccess && !hasMajor && !hasMinor {
p.hasWildCard = true
}
p.insts = append(p.insts, acceptBlock(dev.Allow)...)
// set blockSym to the first instruction we added in this iteration
p.insts[prevBlockLastIdx+1] = p.insts[prevBlockLastIdx+1].Sym(blockSym)
p.blockID++
return nil
}
func (p *program) finalize() (asm.Instructions, error) {
if p.hasWildCard {
// acceptBlock with asm.Return() is already inserted
return p.insts, nil
}
blockSym := fmt.Sprintf("block-%d", p.blockID)
p.insts = append(p.insts,
// R0 <- 0
asm.Mov.Imm32(asm.R0, 0).Sym(blockSym),
asm.Return(),
)
p.blockID = -1
return p.insts, nil
}
func acceptBlock(accept bool) asm.Instructions {
v := int32(0)
if accept {
v = 1
}
return []asm.Instruction{
// R0 <- v
asm.Mov.Imm32(asm.R0, v),
asm.Return(),
}
}
package ebpf
import (
"github.com/cilium/ebpf"
"github.com/cilium/ebpf/asm"
"github.com/pkg/errors"
"golang.org/x/sys/unix"
)
// LoadAttachCgroupDeviceFilter installs eBPF device filter program to /sys/fs/cgroup/<foo> directory.
//
// Requires the system to be running in cgroup2 unified-mode with kernel >= 4.15 .
//
// https://github.com/torvalds/linux/commit/ebc614f687369f9df99828572b1d85a7c2de3d92
func LoadAttachCgroupDeviceFilter(insts asm.Instructions, license string, dirFD int) (func() error, error) {
nilCloser := func() error {
return nil
}
// Increase `ulimit -l` limit to avoid BPF_PROG_LOAD error (#2167).
// This limit is not inherited into the container.
memlockLimit := &unix.Rlimit{
Cur: unix.RLIM_INFINITY,
Max: unix.RLIM_INFINITY,
}
_ = unix.Setrlimit(unix.RLIMIT_MEMLOCK, memlockLimit)
spec := &ebpf.ProgramSpec{
Type: ebpf.CGroupDevice,
Instructions: insts,
License: license,
}
prog, err := ebpf.NewProgram(spec)
if err != nil {
return nilCloser, err
}
if err := prog.Attach(dirFD, ebpf.AttachCGroupDevice, unix.BPF_F_ALLOW_MULTI); err != nil {
return nilCloser, errors.Wrap(err, "failed to call BPF_PROG_ATTACH (BPF_CGROUP_DEVICE, BPF_F_ALLOW_MULTI)")
}
closer := func() error {
if err := prog.Detach(dirFD, ebpf.AttachCGroupDevice, unix.BPF_F_ALLOW_MULTI); err != nil {
return errors.Wrap(err, "failed to call BPF_PROG_DETACH (BPF_CGROUP_DEVICE, BPF_F_ALLOW_MULTI)")
}
return nil
}
return closer, nil
}
...@@ -5,7 +5,6 @@ package fs ...@@ -5,7 +5,6 @@ package fs
import ( import (
"fmt" "fmt"
"io" "io"
"io/ioutil"
"os" "os"
"path/filepath" "path/filepath"
"sync" "sync"
...@@ -33,14 +32,6 @@ var ( ...@@ -33,14 +32,6 @@ var (
&FreezerGroup{}, &FreezerGroup{},
&NameGroup{GroupName: "name=systemd", Join: true}, &NameGroup{GroupName: "name=systemd", Join: true},
} }
subsystemsUnified = subsystemSet{
&CpusetGroupV2{},
&FreezerGroupV2{},
&CpuGroupV2{},
&MemoryGroupV2{},
&IOGroupV2{},
&PidsGroupV2{},
}
HugePageSizes, _ = cgroups.GetHugePageSize() HugePageSizes, _ = cgroups.GetHugePageSize()
) )
...@@ -138,9 +129,6 @@ func isIgnorableError(rootless bool, err error) bool { ...@@ -138,9 +129,6 @@ func isIgnorableError(rootless bool, err error) bool {
} }
func (m *Manager) getSubsystems() subsystemSet { func (m *Manager) getSubsystems() subsystemSet {
if cgroups.IsCgroup2UnifiedMode() {
return subsystemsUnified
}
return subsystemsLegacy return subsystemsLegacy
} }
...@@ -224,6 +212,10 @@ func (m *Manager) GetPaths() map[string]string { ...@@ -224,6 +212,10 @@ func (m *Manager) GetPaths() map[string]string {
return paths return paths
} }
func (m *Manager) GetUnifiedPath() (string, error) {
return "", errors.New("unified path is only supported when running in unified mode")
}
func (m *Manager) GetStats() (*cgroups.Stats, error) { func (m *Manager) GetStats() (*cgroups.Stats, error) {
m.mu.Lock() m.mu.Lock()
defer m.mu.Unlock() defer m.mu.Unlock()
...@@ -377,23 +369,6 @@ func (raw *cgroupData) join(subsystem string) (string, error) { ...@@ -377,23 +369,6 @@ func (raw *cgroupData) join(subsystem string) (string, error) {
return path, nil return path, nil
} }
func writeFile(dir, file, data string) error {
// Normally dir should not be empty, one case is that cgroup subsystem
// is not mounted, we will get empty dir, and we want it fail here.
if dir == "" {
return fmt.Errorf("no such directory for %s", file)
}
if err := ioutil.WriteFile(filepath.Join(dir, file), []byte(data), 0700); err != nil {
return fmt.Errorf("failed to write %v to %v: %v", data, file, err)
}
return nil
}
func readFile(dir, file string) (string, error) {
data, err := ioutil.ReadFile(filepath.Join(dir, file))
return string(data), err
}
func removePath(p string, err error) error { func removePath(p string, err error) error {
if err != nil { if err != nil {
return err return err
...@@ -430,3 +405,7 @@ func CheckCpushares(path string, c uint64) error { ...@@ -430,3 +405,7 @@ func CheckCpushares(path string, c uint64) error {
return nil return nil
} }
func (m *Manager) GetCgroups() (*configs.Cgroup, error) {
return m.Cgroups, nil
}
...@@ -11,6 +11,7 @@ import ( ...@@ -11,6 +11,7 @@ import (
"strings" "strings"
"github.com/opencontainers/runc/libcontainer/cgroups" "github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/cgroups/fscommon"
"github.com/opencontainers/runc/libcontainer/configs" "github.com/opencontainers/runc/libcontainer/configs"
) )
...@@ -31,41 +32,41 @@ func (s *BlkioGroup) Apply(d *cgroupData) error { ...@@ -31,41 +32,41 @@ func (s *BlkioGroup) Apply(d *cgroupData) error {
func (s *BlkioGroup) Set(path string, cgroup *configs.Cgroup) error { func (s *BlkioGroup) Set(path string, cgroup *configs.Cgroup) error {
if cgroup.Resources.BlkioWeight != 0 { if cgroup.Resources.BlkioWeight != 0 {
if err := writeFile(path, "blkio.weight", strconv.FormatUint(uint64(cgroup.Resources.BlkioWeight), 10)); err != nil { if err := fscommon.WriteFile(path, "blkio.weight", strconv.FormatUint(uint64(cgroup.Resources.BlkioWeight), 10)); err != nil {
return err return err
} }
} }
if cgroup.Resources.BlkioLeafWeight != 0 { if cgroup.Resources.BlkioLeafWeight != 0 {
if err := writeFile(path, "blkio.leaf_weight", strconv.FormatUint(uint64(cgroup.Resources.BlkioLeafWeight), 10)); err != nil { if err := fscommon.WriteFile(path, "blkio.leaf_weight", strconv.FormatUint(uint64(cgroup.Resources.BlkioLeafWeight), 10)); err != nil {
return err return err
} }
} }
for _, wd := range cgroup.Resources.BlkioWeightDevice { for _, wd := range cgroup.Resources.BlkioWeightDevice {
if err := writeFile(path, "blkio.weight_device", wd.WeightString()); err != nil { if err := fscommon.WriteFile(path, "blkio.weight_device", wd.WeightString()); err != nil {
return err return err
} }
if err := writeFile(path, "blkio.leaf_weight_device", wd.LeafWeightString()); err != nil { if err := fscommon.WriteFile(path, "blkio.leaf_weight_device", wd.LeafWeightString()); err != nil {
return err return err
} }
} }
for _, td := range cgroup.Resources.BlkioThrottleReadBpsDevice { for _, td := range cgroup.Resources.BlkioThrottleReadBpsDevice {
if err := writeFile(path, "blkio.throttle.read_bps_device", td.String()); err != nil { if err := fscommon.WriteFile(path, "blkio.throttle.read_bps_device", td.String()); err != nil {
return err return err
} }
} }
for _, td := range cgroup.Resources.BlkioThrottleWriteBpsDevice { for _, td := range cgroup.Resources.BlkioThrottleWriteBpsDevice {
if err := writeFile(path, "blkio.throttle.write_bps_device", td.String()); err != nil { if err := fscommon.WriteFile(path, "blkio.throttle.write_bps_device", td.String()); err != nil {
return err return err
} }
} }
for _, td := range cgroup.Resources.BlkioThrottleReadIOPSDevice { for _, td := range cgroup.Resources.BlkioThrottleReadIOPSDevice {
if err := writeFile(path, "blkio.throttle.read_iops_device", td.String()); err != nil { if err := fscommon.WriteFile(path, "blkio.throttle.read_iops_device", td.String()); err != nil {
return err return err
} }
} }
for _, td := range cgroup.Resources.BlkioThrottleWriteIOPSDevice { for _, td := range cgroup.Resources.BlkioThrottleWriteIOPSDevice {
if err := writeFile(path, "blkio.throttle.write_iops_device", td.String()); err != nil { if err := fscommon.WriteFile(path, "blkio.throttle.write_iops_device", td.String()); err != nil {
return err return err
} }
} }
......
...@@ -9,6 +9,7 @@ import ( ...@@ -9,6 +9,7 @@ import (
"strconv" "strconv"
"github.com/opencontainers/runc/libcontainer/cgroups" "github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/cgroups/fscommon"
"github.com/opencontainers/runc/libcontainer/configs" "github.com/opencontainers/runc/libcontainer/configs"
) )
...@@ -51,12 +52,12 @@ func (s *CpuGroup) ApplyDir(path string, cgroup *configs.Cgroup, pid int) error ...@@ -51,12 +52,12 @@ func (s *CpuGroup) ApplyDir(path string, cgroup *configs.Cgroup, pid int) error
func (s *CpuGroup) SetRtSched(path string, cgroup *configs.Cgroup) error { func (s *CpuGroup) SetRtSched(path string, cgroup *configs.Cgroup) error {
if cgroup.Resources.CpuRtPeriod != 0 { if cgroup.Resources.CpuRtPeriod != 0 {
if err := writeFile(path, "cpu.rt_period_us", strconv.FormatUint(cgroup.Resources.CpuRtPeriod, 10)); err != nil { if err := fscommon.WriteFile(path, "cpu.rt_period_us", strconv.FormatUint(cgroup.Resources.CpuRtPeriod, 10)); err != nil {
return err return err
} }
} }
if cgroup.Resources.CpuRtRuntime != 0 { if cgroup.Resources.CpuRtRuntime != 0 {
if err := writeFile(path, "cpu.rt_runtime_us", strconv.FormatInt(cgroup.Resources.CpuRtRuntime, 10)); err != nil { if err := fscommon.WriteFile(path, "cpu.rt_runtime_us", strconv.FormatInt(cgroup.Resources.CpuRtRuntime, 10)); err != nil {
return err return err
} }
} }
...@@ -65,17 +66,17 @@ func (s *CpuGroup) SetRtSched(path string, cgroup *configs.Cgroup) error { ...@@ -65,17 +66,17 @@ func (s *CpuGroup) SetRtSched(path string, cgroup *configs.Cgroup) error {
func (s *CpuGroup) Set(path string, cgroup *configs.Cgroup) error { func (s *CpuGroup) Set(path string, cgroup *configs.Cgroup) error {
if cgroup.Resources.CpuShares != 0 { if cgroup.Resources.CpuShares != 0 {
if err := writeFile(path, "cpu.shares", strconv.FormatUint(cgroup.Resources.CpuShares, 10)); err != nil { if err := fscommon.WriteFile(path, "cpu.shares", strconv.FormatUint(cgroup.Resources.CpuShares, 10)); err != nil {
return err return err
} }
} }
if cgroup.Resources.CpuPeriod != 0 { if cgroup.Resources.CpuPeriod != 0 {
if err := writeFile(path, "cpu.cfs_period_us", strconv.FormatUint(cgroup.Resources.CpuPeriod, 10)); err != nil { if err := fscommon.WriteFile(path, "cpu.cfs_period_us", strconv.FormatUint(cgroup.Resources.CpuPeriod, 10)); err != nil {
return err return err
} }
} }
if cgroup.Resources.CpuQuota != 0 { if cgroup.Resources.CpuQuota != 0 {
if err := writeFile(path, "cpu.cfs_quota_us", strconv.FormatInt(cgroup.Resources.CpuQuota, 10)); err != nil { if err := fscommon.WriteFile(path, "cpu.cfs_quota_us", strconv.FormatInt(cgroup.Resources.CpuQuota, 10)); err != nil {
return err return err
} }
} }
...@@ -98,7 +99,7 @@ func (s *CpuGroup) GetStats(path string, stats *cgroups.Stats) error { ...@@ -98,7 +99,7 @@ func (s *CpuGroup) GetStats(path string, stats *cgroups.Stats) error {
sc := bufio.NewScanner(f) sc := bufio.NewScanner(f)
for sc.Scan() { for sc.Scan() {
t, v, err := getCgroupParamKeyValue(sc.Text()) t, v, err := fscommon.GetCgroupParamKeyValue(sc.Text())
if err != nil { if err != nil {
return err return err
} }
......
...@@ -10,6 +10,7 @@ import ( ...@@ -10,6 +10,7 @@ import (
"strings" "strings"
"github.com/opencontainers/runc/libcontainer/cgroups" "github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/cgroups/fscommon"
"github.com/opencontainers/runc/libcontainer/configs" "github.com/opencontainers/runc/libcontainer/configs"
"github.com/opencontainers/runc/libcontainer/system" "github.com/opencontainers/runc/libcontainer/system"
) )
...@@ -51,7 +52,7 @@ func (s *CpuacctGroup) GetStats(path string, stats *cgroups.Stats) error { ...@@ -51,7 +52,7 @@ func (s *CpuacctGroup) GetStats(path string, stats *cgroups.Stats) error {
return err return err
} }
totalUsage, err := getCgroupParamUint(path, "cpuacct.usage") totalUsage, err := fscommon.GetCgroupParamUint(path, "cpuacct.usage")
if err != nil { if err != nil {
return err return err
} }
...@@ -85,8 +86,8 @@ func getCpuUsageBreakdown(path string) (uint64, uint64, error) { ...@@ -85,8 +86,8 @@ func getCpuUsageBreakdown(path string) (uint64, uint64, error) {
return 0, 0, err return 0, 0, err
} }
fields := strings.Fields(string(data)) fields := strings.Fields(string(data))
if len(fields) != 4 { if len(fields) < 4 {
return 0, 0, fmt.Errorf("failure - %s is expected to have 4 fields", filepath.Join(path, cgroupCpuacctStat)) return 0, 0, fmt.Errorf("failure - %s is expected to have at least 4 fields", filepath.Join(path, cgroupCpuacctStat))
} }
if fields[0] != userField { if fields[0] != userField {
return 0, 0, fmt.Errorf("unexpected field %q in %q, expected %q", fields[0], cgroupCpuacctStat, userField) return 0, 0, fmt.Errorf("unexpected field %q in %q, expected %q", fields[0], cgroupCpuacctStat, userField)
......
...@@ -10,6 +10,7 @@ import ( ...@@ -10,6 +10,7 @@ import (
"path/filepath" "path/filepath"
"github.com/opencontainers/runc/libcontainer/cgroups" "github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/cgroups/fscommon"
"github.com/opencontainers/runc/libcontainer/configs" "github.com/opencontainers/runc/libcontainer/configs"
libcontainerUtils "github.com/opencontainers/runc/libcontainer/utils" libcontainerUtils "github.com/opencontainers/runc/libcontainer/utils"
) )
...@@ -31,12 +32,12 @@ func (s *CpusetGroup) Apply(d *cgroupData) error { ...@@ -31,12 +32,12 @@ func (s *CpusetGroup) Apply(d *cgroupData) error {
func (s *CpusetGroup) Set(path string, cgroup *configs.Cgroup) error { func (s *CpusetGroup) Set(path string, cgroup *configs.Cgroup) error {
if cgroup.Resources.CpusetCpus != "" { if cgroup.Resources.CpusetCpus != "" {
if err := writeFile(path, "cpuset.cpus", cgroup.Resources.CpusetCpus); err != nil { if err := fscommon.WriteFile(path, "cpuset.cpus", cgroup.Resources.CpusetCpus); err != nil {
return err return err
} }
} }
if cgroup.Resources.CpusetMems != "" { if cgroup.Resources.CpusetMems != "" {
if err := writeFile(path, "cpuset.mems", cgroup.Resources.CpusetMems); err != nil { if err := fscommon.WriteFile(path, "cpuset.mems", cgroup.Resources.CpusetMems); err != nil {
return err return err
} }
} }
...@@ -135,12 +136,12 @@ func (s *CpusetGroup) copyIfNeeded(current, parent string) error { ...@@ -135,12 +136,12 @@ func (s *CpusetGroup) copyIfNeeded(current, parent string) error {
} }
if s.isEmpty(currentCpus) { if s.isEmpty(currentCpus) {
if err := writeFile(current, "cpuset.cpus", string(parentCpus)); err != nil { if err := fscommon.WriteFile(current, "cpuset.cpus", string(parentCpus)); err != nil {
return err return err
} }
} }
if s.isEmpty(currentMems) { if s.isEmpty(currentMems) {
if err := writeFile(current, "cpuset.mems", string(parentMems)); err != nil { if err := fscommon.WriteFile(current, "cpuset.mems", string(parentMems)); err != nil {
return err return err
} }
} }
......
// +build linux
package fs
import (
"bytes"
"fmt"
"io/ioutil"
"os"
"path/filepath"
"github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/configs"
libcontainerUtils "github.com/opencontainers/runc/libcontainer/utils"
)
type CpusetGroupV2 struct {
}
func (s *CpusetGroupV2) Name() string {
return "cpuset"
}
func (s *CpusetGroupV2) Apply(d *cgroupData) error {
dir, err := d.path("cpuset")
if err != nil && !cgroups.IsNotFound(err) {
return err
}
return s.ApplyDir(dir, d.config, d.pid)
}
func (s *CpusetGroupV2) Set(path string, cgroup *configs.Cgroup) error {
if cgroup.Resources.CpusetCpus != "" {
if err := writeFile(path, "cpuset.cpus", cgroup.Resources.CpusetCpus); err != nil {
return err
}
}
if cgroup.Resources.CpusetMems != "" {
if err := writeFile(path, "cpuset.mems", cgroup.Resources.CpusetMems); err != nil {
return err
}
}
return nil
}
func (s *CpusetGroupV2) Remove(d *cgroupData) error {
return removePath(d.path("cpuset"))
}
func (s *CpusetGroupV2) GetStats(path string, stats *cgroups.Stats) error {
return nil
}
func (s *CpusetGroupV2) ApplyDir(dir string, cgroup *configs.Cgroup, pid int) error {
// This might happen if we have no cpuset cgroup mounted.
// Just do nothing and don't fail.
if dir == "" {
return nil
}
mountInfo, err := ioutil.ReadFile("/proc/self/mountinfo")
if err != nil {
return err
}
root := filepath.Dir(cgroups.GetClosestMountpointAncestor(dir, string(mountInfo)))
// 'ensureParent' start with parent because we don't want to
// explicitly inherit from parent, it could conflict with
// 'cpuset.cpu_exclusive'.
if err := s.ensureParent(filepath.Dir(dir), root); err != nil {
return err
}
if err := os.MkdirAll(dir, 0755); err != nil {
return err
}
// We didn't inherit cpuset configs from parent, but we have
// to ensure cpuset configs are set before moving task into the
// cgroup.
// The logic is, if user specified cpuset configs, use these
// specified configs, otherwise, inherit from parent. This makes
// cpuset configs work correctly with 'cpuset.cpu_exclusive', and
// keep backward compatibility.
if err := s.ensureCpusAndMems(dir, cgroup); err != nil {
return err
}
// because we are not using d.join we need to place the pid into the procs file
// unlike the other subsystems
return cgroups.WriteCgroupProc(dir, pid)
}
func (s *CpusetGroupV2) getSubsystemSettings(parent string) (cpus []byte, mems []byte, err error) {
if cpus, err = ioutil.ReadFile(filepath.Join(parent, "cpuset.cpus.effective")); err != nil {
return
}
if mems, err = ioutil.ReadFile(filepath.Join(parent, "cpuset.mems.effective")); err != nil {
return
}
return cpus, mems, nil
}
// ensureParent makes sure that the parent directory of current is created
// and populated with the proper cpus and mems files copied from
// it's parent.
func (s *CpusetGroupV2) ensureParent(current, root string) error {
parent := filepath.Dir(current)
if libcontainerUtils.CleanPath(parent) == root {
return nil
}
// Avoid infinite recursion.
if parent == current {
return fmt.Errorf("cpuset: cgroup parent path outside cgroup root")
}
if err := s.ensureParent(parent, root); err != nil {
return err
}
if err := os.MkdirAll(current, 0755); err != nil {
return err
}
return s.copyIfNeeded(current, parent)
}
// copyIfNeeded copies the cpuset.cpus and cpuset.mems from the parent
// directory to the current directory if the file's contents are 0
func (s *CpusetGroupV2) copyIfNeeded(current, parent string) error {
var (
err error
currentCpus, currentMems []byte
parentCpus, parentMems []byte
)
if currentCpus, currentMems, err = s.getSubsystemSettings(current); err != nil {
return err
}
if parentCpus, parentMems, err = s.getSubsystemSettings(parent); err != nil {
return err
}
if s.isEmpty(currentCpus) {
if err := writeFile(current, "cpuset.cpus", string(parentCpus)); err != nil {
return err
}
}
if s.isEmpty(currentMems) {
if err := writeFile(current, "cpuset.mems", string(parentMems)); err != nil {
return err
}
}
return nil
}
func (s *CpusetGroupV2) isEmpty(b []byte) bool {
return len(bytes.Trim(b, "\n")) == 0
}
func (s *CpusetGroupV2) ensureCpusAndMems(path string, cgroup *configs.Cgroup) error {
if err := s.Set(path, cgroup); err != nil {
return err
}
return s.copyIfNeeded(path, filepath.Dir(path))
}
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment