syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
// Copyright 2019 The Go Authors. All rights reserved.
|
|
|
|
// Use of this source code is governed by a BSD-style
|
|
|
|
// license that can be found in the LICENSE file.
|
|
|
|
|
|
|
|
// +build linux,cgo
|
|
|
|
|
|
|
|
package cgotest
|
|
|
|
|
|
|
|
import (
|
|
|
|
"fmt"
|
2021-04-03 02:10:47 -06:00
|
|
|
"os"
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
"strings"
|
|
|
|
"syscall"
|
|
|
|
"testing"
|
|
|
|
)
|
|
|
|
|
|
|
|
// #include <stdio.h>
|
|
|
|
// #include <stdlib.h>
|
|
|
|
// #include <pthread.h>
|
|
|
|
// #include <unistd.h>
|
|
|
|
// #include <sys/types.h>
|
|
|
|
//
|
|
|
|
// pthread_t *t = NULL;
|
|
|
|
// pthread_mutex_t mu;
|
|
|
|
// int nts = 0;
|
|
|
|
// int all_done = 0;
|
|
|
|
//
|
|
|
|
// static void *aFn(void *vargp) {
|
|
|
|
// int done = 0;
|
|
|
|
// while (!done) {
|
|
|
|
// usleep(100);
|
|
|
|
// pthread_mutex_lock(&mu);
|
|
|
|
// done = all_done;
|
|
|
|
// pthread_mutex_unlock(&mu);
|
|
|
|
// }
|
|
|
|
// return NULL;
|
|
|
|
// }
|
|
|
|
//
|
|
|
|
// void trial(int argc) {
|
|
|
|
// int i;
|
|
|
|
// nts = argc;
|
|
|
|
// t = calloc(nts, sizeof(pthread_t));
|
|
|
|
// pthread_mutex_init(&mu, NULL);
|
|
|
|
// for (i = 0; i < nts; i++) {
|
|
|
|
// pthread_create(&t[i], NULL, aFn, NULL);
|
|
|
|
// }
|
|
|
|
// }
|
|
|
|
//
|
|
|
|
// void cleanup(void) {
|
|
|
|
// int i;
|
|
|
|
// pthread_mutex_lock(&mu);
|
|
|
|
// all_done = 1;
|
|
|
|
// pthread_mutex_unlock(&mu);
|
|
|
|
// for (i = 0; i < nts; i++) {
|
|
|
|
// pthread_join(t[i], NULL);
|
|
|
|
// }
|
|
|
|
// pthread_mutex_destroy(&mu);
|
|
|
|
// free(t);
|
|
|
|
// }
|
|
|
|
import "C"
|
|
|
|
|
|
|
|
// compareStatus is used to confirm the contents of the thread
|
|
|
|
// specific status files match expectations.
|
|
|
|
func compareStatus(filter, expect string) error {
|
syscall: improve TestSetuidEtc() /proc/ parsing against races
TestSetuidEtc() was failing sporadically on linux-ppc64. From the
three https://build.golang.org/ logs, it looked like the logged
errors could be associated with threads dying, but proc reads
were, in some way, racing with their demise.
Exploring ways to increase thread demise, revealed that races
of this type can happen on non-ppc64 systems, and that
os.IsNotExist(err) was not a sufficient error condition test
for a thread's status file disappearing. This change includes a
fix for that to.
The actual issue on linux-ppc64 appears to be tied to PID reaping
and reuse latency on whatever the build test environment is for
linux-ppc64-buildlet. I suspect this can happen on any linux
system, however, especially where the container has a limited PID
range.
The fix for this, limited to the test (the runtime syscall support
is unchanged), is to confirm that the Pid for the interrogated
thread's /proc/<TID>/status file confirms that it is still
associated with the test-process' PID.
linux-ppc64-buildlet:
go/bin/go test syscall -run=TestSetuidEtc -count=10000
ok syscall 104.285s
Fixes #42462
Change-Id: I55c84ab8361003570a405fa52ffec4949bf91113
Reviewed-on: https://go-review.googlesource.com/c/go/+/268717
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Tobias Klauser <tobias.klauser@gmail.com>
2020-11-09 20:28:04 -07:00
|
|
|
expected := filter + expect
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
pid := syscall.Getpid()
|
2021-04-03 02:10:47 -06:00
|
|
|
fs, err := os.ReadDir(fmt.Sprintf("/proc/%d/task", pid))
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
if err != nil {
|
|
|
|
return fmt.Errorf("unable to find %d tasks: %v", pid, err)
|
|
|
|
}
|
syscall: improve TestSetuidEtc() /proc/ parsing against races
TestSetuidEtc() was failing sporadically on linux-ppc64. From the
three https://build.golang.org/ logs, it looked like the logged
errors could be associated with threads dying, but proc reads
were, in some way, racing with their demise.
Exploring ways to increase thread demise, revealed that races
of this type can happen on non-ppc64 systems, and that
os.IsNotExist(err) was not a sufficient error condition test
for a thread's status file disappearing. This change includes a
fix for that to.
The actual issue on linux-ppc64 appears to be tied to PID reaping
and reuse latency on whatever the build test environment is for
linux-ppc64-buildlet. I suspect this can happen on any linux
system, however, especially where the container has a limited PID
range.
The fix for this, limited to the test (the runtime syscall support
is unchanged), is to confirm that the Pid for the interrogated
thread's /proc/<TID>/status file confirms that it is still
associated with the test-process' PID.
linux-ppc64-buildlet:
go/bin/go test syscall -run=TestSetuidEtc -count=10000
ok syscall 104.285s
Fixes #42462
Change-Id: I55c84ab8361003570a405fa52ffec4949bf91113
Reviewed-on: https://go-review.googlesource.com/c/go/+/268717
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Tobias Klauser <tobias.klauser@gmail.com>
2020-11-09 20:28:04 -07:00
|
|
|
expectedProc := fmt.Sprintf("Pid:\t%d", pid)
|
|
|
|
foundAThread := false
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
for _, f := range fs {
|
|
|
|
tf := fmt.Sprintf("/proc/%s/status", f.Name())
|
2021-04-03 02:10:47 -06:00
|
|
|
d, err := os.ReadFile(tf)
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
if err != nil {
|
syscall: improve TestSetuidEtc() /proc/ parsing against races
TestSetuidEtc() was failing sporadically on linux-ppc64. From the
three https://build.golang.org/ logs, it looked like the logged
errors could be associated with threads dying, but proc reads
were, in some way, racing with their demise.
Exploring ways to increase thread demise, revealed that races
of this type can happen on non-ppc64 systems, and that
os.IsNotExist(err) was not a sufficient error condition test
for a thread's status file disappearing. This change includes a
fix for that to.
The actual issue on linux-ppc64 appears to be tied to PID reaping
and reuse latency on whatever the build test environment is for
linux-ppc64-buildlet. I suspect this can happen on any linux
system, however, especially where the container has a limited PID
range.
The fix for this, limited to the test (the runtime syscall support
is unchanged), is to confirm that the Pid for the interrogated
thread's /proc/<TID>/status file confirms that it is still
associated with the test-process' PID.
linux-ppc64-buildlet:
go/bin/go test syscall -run=TestSetuidEtc -count=10000
ok syscall 104.285s
Fixes #42462
Change-Id: I55c84ab8361003570a405fa52ffec4949bf91113
Reviewed-on: https://go-review.googlesource.com/c/go/+/268717
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Tobias Klauser <tobias.klauser@gmail.com>
2020-11-09 20:28:04 -07:00
|
|
|
// There are a surprising number of ways this
|
|
|
|
// can error out on linux. We've seen all of
|
|
|
|
// the following, so treat any error here as
|
|
|
|
// equivalent to the "process is gone":
|
|
|
|
// os.IsNotExist(err),
|
|
|
|
// "... : no such process",
|
|
|
|
// "... : bad file descriptor.
|
|
|
|
continue
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
}
|
|
|
|
lines := strings.Split(string(d), "\n")
|
|
|
|
for _, line := range lines {
|
syscall: improve TestSetuidEtc() /proc/ parsing against races
TestSetuidEtc() was failing sporadically on linux-ppc64. From the
three https://build.golang.org/ logs, it looked like the logged
errors could be associated with threads dying, but proc reads
were, in some way, racing with their demise.
Exploring ways to increase thread demise, revealed that races
of this type can happen on non-ppc64 systems, and that
os.IsNotExist(err) was not a sufficient error condition test
for a thread's status file disappearing. This change includes a
fix for that to.
The actual issue on linux-ppc64 appears to be tied to PID reaping
and reuse latency on whatever the build test environment is for
linux-ppc64-buildlet. I suspect this can happen on any linux
system, however, especially where the container has a limited PID
range.
The fix for this, limited to the test (the runtime syscall support
is unchanged), is to confirm that the Pid for the interrogated
thread's /proc/<TID>/status file confirms that it is still
associated with the test-process' PID.
linux-ppc64-buildlet:
go/bin/go test syscall -run=TestSetuidEtc -count=10000
ok syscall 104.285s
Fixes #42462
Change-Id: I55c84ab8361003570a405fa52ffec4949bf91113
Reviewed-on: https://go-review.googlesource.com/c/go/+/268717
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Tobias Klauser <tobias.klauser@gmail.com>
2020-11-09 20:28:04 -07:00
|
|
|
// Different kernel vintages pad differently.
|
|
|
|
line = strings.TrimSpace(line)
|
|
|
|
if strings.HasPrefix(line, "Pid:\t") {
|
|
|
|
// On loaded systems, it is possible
|
|
|
|
// for a TID to be reused really
|
|
|
|
// quickly. As such, we need to
|
|
|
|
// validate that the thread status
|
|
|
|
// info we just read is a task of the
|
|
|
|
// same process PID as we are
|
|
|
|
// currently running, and not a
|
|
|
|
// recently terminated thread
|
|
|
|
// resurfaced in a different process.
|
|
|
|
if line != expectedProc {
|
|
|
|
break
|
|
|
|
}
|
|
|
|
// Fall through in the unlikely case
|
|
|
|
// that filter at some point is
|
|
|
|
// "Pid:\t".
|
|
|
|
}
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
if strings.HasPrefix(line, filter) {
|
|
|
|
if line != expected {
|
syscall: improve TestSetuidEtc() /proc/ parsing against races
TestSetuidEtc() was failing sporadically on linux-ppc64. From the
three https://build.golang.org/ logs, it looked like the logged
errors could be associated with threads dying, but proc reads
were, in some way, racing with their demise.
Exploring ways to increase thread demise, revealed that races
of this type can happen on non-ppc64 systems, and that
os.IsNotExist(err) was not a sufficient error condition test
for a thread's status file disappearing. This change includes a
fix for that to.
The actual issue on linux-ppc64 appears to be tied to PID reaping
and reuse latency on whatever the build test environment is for
linux-ppc64-buildlet. I suspect this can happen on any linux
system, however, especially where the container has a limited PID
range.
The fix for this, limited to the test (the runtime syscall support
is unchanged), is to confirm that the Pid for the interrogated
thread's /proc/<TID>/status file confirms that it is still
associated with the test-process' PID.
linux-ppc64-buildlet:
go/bin/go test syscall -run=TestSetuidEtc -count=10000
ok syscall 104.285s
Fixes #42462
Change-Id: I55c84ab8361003570a405fa52ffec4949bf91113
Reviewed-on: https://go-review.googlesource.com/c/go/+/268717
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Tobias Klauser <tobias.klauser@gmail.com>
2020-11-09 20:28:04 -07:00
|
|
|
return fmt.Errorf("%q got:%q want:%q (bad) [pid=%d file:'%s' %v]\n", tf, line, expected, pid, string(d), expectedProc)
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
}
|
syscall: improve TestSetuidEtc() /proc/ parsing against races
TestSetuidEtc() was failing sporadically on linux-ppc64. From the
three https://build.golang.org/ logs, it looked like the logged
errors could be associated with threads dying, but proc reads
were, in some way, racing with their demise.
Exploring ways to increase thread demise, revealed that races
of this type can happen on non-ppc64 systems, and that
os.IsNotExist(err) was not a sufficient error condition test
for a thread's status file disappearing. This change includes a
fix for that to.
The actual issue on linux-ppc64 appears to be tied to PID reaping
and reuse latency on whatever the build test environment is for
linux-ppc64-buildlet. I suspect this can happen on any linux
system, however, especially where the container has a limited PID
range.
The fix for this, limited to the test (the runtime syscall support
is unchanged), is to confirm that the Pid for the interrogated
thread's /proc/<TID>/status file confirms that it is still
associated with the test-process' PID.
linux-ppc64-buildlet:
go/bin/go test syscall -run=TestSetuidEtc -count=10000
ok syscall 104.285s
Fixes #42462
Change-Id: I55c84ab8361003570a405fa52ffec4949bf91113
Reviewed-on: https://go-review.googlesource.com/c/go/+/268717
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Tobias Klauser <tobias.klauser@gmail.com>
2020-11-09 20:28:04 -07:00
|
|
|
foundAThread = true
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
break
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
syscall: improve TestSetuidEtc() /proc/ parsing against races
TestSetuidEtc() was failing sporadically on linux-ppc64. From the
three https://build.golang.org/ logs, it looked like the logged
errors could be associated with threads dying, but proc reads
were, in some way, racing with their demise.
Exploring ways to increase thread demise, revealed that races
of this type can happen on non-ppc64 systems, and that
os.IsNotExist(err) was not a sufficient error condition test
for a thread's status file disappearing. This change includes a
fix for that to.
The actual issue on linux-ppc64 appears to be tied to PID reaping
and reuse latency on whatever the build test environment is for
linux-ppc64-buildlet. I suspect this can happen on any linux
system, however, especially where the container has a limited PID
range.
The fix for this, limited to the test (the runtime syscall support
is unchanged), is to confirm that the Pid for the interrogated
thread's /proc/<TID>/status file confirms that it is still
associated with the test-process' PID.
linux-ppc64-buildlet:
go/bin/go test syscall -run=TestSetuidEtc -count=10000
ok syscall 104.285s
Fixes #42462
Change-Id: I55c84ab8361003570a405fa52ffec4949bf91113
Reviewed-on: https://go-review.googlesource.com/c/go/+/268717
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Tobias Klauser <tobias.klauser@gmail.com>
2020-11-09 20:28:04 -07:00
|
|
|
if !foundAThread {
|
|
|
|
return fmt.Errorf("found no thread /proc/<TID>/status files for process %q", expectedProc)
|
|
|
|
}
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
return nil
|
|
|
|
}
|
|
|
|
|
|
|
|
// test1435 test 9 glibc implemented setuid/gid syscall functions are
|
|
|
|
// mapped. This test is a slightly more expansive test than that of
|
|
|
|
// src/syscall/syscall_linux_test.go:TestSetuidEtc() insofar as it
|
|
|
|
// launches concurrent threads from C code via CGo and validates that
|
|
|
|
// they are subject to the system calls being tested. For the actual
|
|
|
|
// Go functionality being tested here, the syscall_linux_test version
|
|
|
|
// is considered authoritative, but non-trivial improvements to that
|
|
|
|
// should be mirrored here.
|
|
|
|
func test1435(t *testing.T) {
|
|
|
|
if syscall.Getuid() != 0 {
|
|
|
|
t.Skip("skipping root only test")
|
|
|
|
}
|
|
|
|
|
|
|
|
// Launch some threads in C.
|
|
|
|
const cts = 5
|
|
|
|
C.trial(cts)
|
|
|
|
defer C.cleanup()
|
|
|
|
|
|
|
|
vs := []struct {
|
|
|
|
call string
|
|
|
|
fn func() error
|
|
|
|
filter, expect string
|
|
|
|
}{
|
syscall: improve TestSetuidEtc() /proc/ parsing against races
TestSetuidEtc() was failing sporadically on linux-ppc64. From the
three https://build.golang.org/ logs, it looked like the logged
errors could be associated with threads dying, but proc reads
were, in some way, racing with their demise.
Exploring ways to increase thread demise, revealed that races
of this type can happen on non-ppc64 systems, and that
os.IsNotExist(err) was not a sufficient error condition test
for a thread's status file disappearing. This change includes a
fix for that to.
The actual issue on linux-ppc64 appears to be tied to PID reaping
and reuse latency on whatever the build test environment is for
linux-ppc64-buildlet. I suspect this can happen on any linux
system, however, especially where the container has a limited PID
range.
The fix for this, limited to the test (the runtime syscall support
is unchanged), is to confirm that the Pid for the interrogated
thread's /proc/<TID>/status file confirms that it is still
associated with the test-process' PID.
linux-ppc64-buildlet:
go/bin/go test syscall -run=TestSetuidEtc -count=10000
ok syscall 104.285s
Fixes #42462
Change-Id: I55c84ab8361003570a405fa52ffec4949bf91113
Reviewed-on: https://go-review.googlesource.com/c/go/+/268717
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Tobias Klauser <tobias.klauser@gmail.com>
2020-11-09 20:28:04 -07:00
|
|
|
{call: "Setegid(1)", fn: func() error { return syscall.Setegid(1) }, filter: "Gid:", expect: "\t0\t1\t0\t1"},
|
|
|
|
{call: "Setegid(0)", fn: func() error { return syscall.Setegid(0) }, filter: "Gid:", expect: "\t0\t0\t0\t0"},
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
|
syscall: improve TestSetuidEtc() /proc/ parsing against races
TestSetuidEtc() was failing sporadically on linux-ppc64. From the
three https://build.golang.org/ logs, it looked like the logged
errors could be associated with threads dying, but proc reads
were, in some way, racing with their demise.
Exploring ways to increase thread demise, revealed that races
of this type can happen on non-ppc64 systems, and that
os.IsNotExist(err) was not a sufficient error condition test
for a thread's status file disappearing. This change includes a
fix for that to.
The actual issue on linux-ppc64 appears to be tied to PID reaping
and reuse latency on whatever the build test environment is for
linux-ppc64-buildlet. I suspect this can happen on any linux
system, however, especially where the container has a limited PID
range.
The fix for this, limited to the test (the runtime syscall support
is unchanged), is to confirm that the Pid for the interrogated
thread's /proc/<TID>/status file confirms that it is still
associated with the test-process' PID.
linux-ppc64-buildlet:
go/bin/go test syscall -run=TestSetuidEtc -count=10000
ok syscall 104.285s
Fixes #42462
Change-Id: I55c84ab8361003570a405fa52ffec4949bf91113
Reviewed-on: https://go-review.googlesource.com/c/go/+/268717
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Tobias Klauser <tobias.klauser@gmail.com>
2020-11-09 20:28:04 -07:00
|
|
|
{call: "Seteuid(1)", fn: func() error { return syscall.Seteuid(1) }, filter: "Uid:", expect: "\t0\t1\t0\t1"},
|
|
|
|
{call: "Setuid(0)", fn: func() error { return syscall.Setuid(0) }, filter: "Uid:", expect: "\t0\t0\t0\t0"},
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
|
syscall: improve TestSetuidEtc() /proc/ parsing against races
TestSetuidEtc() was failing sporadically on linux-ppc64. From the
three https://build.golang.org/ logs, it looked like the logged
errors could be associated with threads dying, but proc reads
were, in some way, racing with their demise.
Exploring ways to increase thread demise, revealed that races
of this type can happen on non-ppc64 systems, and that
os.IsNotExist(err) was not a sufficient error condition test
for a thread's status file disappearing. This change includes a
fix for that to.
The actual issue on linux-ppc64 appears to be tied to PID reaping
and reuse latency on whatever the build test environment is for
linux-ppc64-buildlet. I suspect this can happen on any linux
system, however, especially where the container has a limited PID
range.
The fix for this, limited to the test (the runtime syscall support
is unchanged), is to confirm that the Pid for the interrogated
thread's /proc/<TID>/status file confirms that it is still
associated with the test-process' PID.
linux-ppc64-buildlet:
go/bin/go test syscall -run=TestSetuidEtc -count=10000
ok syscall 104.285s
Fixes #42462
Change-Id: I55c84ab8361003570a405fa52ffec4949bf91113
Reviewed-on: https://go-review.googlesource.com/c/go/+/268717
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Tobias Klauser <tobias.klauser@gmail.com>
2020-11-09 20:28:04 -07:00
|
|
|
{call: "Setgid(1)", fn: func() error { return syscall.Setgid(1) }, filter: "Gid:", expect: "\t1\t1\t1\t1"},
|
|
|
|
{call: "Setgid(0)", fn: func() error { return syscall.Setgid(0) }, filter: "Gid:", expect: "\t0\t0\t0\t0"},
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
|
syscall: improve TestSetuidEtc() /proc/ parsing against races
TestSetuidEtc() was failing sporadically on linux-ppc64. From the
three https://build.golang.org/ logs, it looked like the logged
errors could be associated with threads dying, but proc reads
were, in some way, racing with their demise.
Exploring ways to increase thread demise, revealed that races
of this type can happen on non-ppc64 systems, and that
os.IsNotExist(err) was not a sufficient error condition test
for a thread's status file disappearing. This change includes a
fix for that to.
The actual issue on linux-ppc64 appears to be tied to PID reaping
and reuse latency on whatever the build test environment is for
linux-ppc64-buildlet. I suspect this can happen on any linux
system, however, especially where the container has a limited PID
range.
The fix for this, limited to the test (the runtime syscall support
is unchanged), is to confirm that the Pid for the interrogated
thread's /proc/<TID>/status file confirms that it is still
associated with the test-process' PID.
linux-ppc64-buildlet:
go/bin/go test syscall -run=TestSetuidEtc -count=10000
ok syscall 104.285s
Fixes #42462
Change-Id: I55c84ab8361003570a405fa52ffec4949bf91113
Reviewed-on: https://go-review.googlesource.com/c/go/+/268717
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Tobias Klauser <tobias.klauser@gmail.com>
2020-11-09 20:28:04 -07:00
|
|
|
{call: "Setgroups([]int{0,1,2,3})", fn: func() error { return syscall.Setgroups([]int{0, 1, 2, 3}) }, filter: "Groups:", expect: "\t0 1 2 3"},
|
|
|
|
{call: "Setgroups(nil)", fn: func() error { return syscall.Setgroups(nil) }, filter: "Groups:", expect: ""},
|
|
|
|
{call: "Setgroups([]int{0})", fn: func() error { return syscall.Setgroups([]int{0}) }, filter: "Groups:", expect: "\t0"},
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
|
syscall: improve TestSetuidEtc() /proc/ parsing against races
TestSetuidEtc() was failing sporadically on linux-ppc64. From the
three https://build.golang.org/ logs, it looked like the logged
errors could be associated with threads dying, but proc reads
were, in some way, racing with their demise.
Exploring ways to increase thread demise, revealed that races
of this type can happen on non-ppc64 systems, and that
os.IsNotExist(err) was not a sufficient error condition test
for a thread's status file disappearing. This change includes a
fix for that to.
The actual issue on linux-ppc64 appears to be tied to PID reaping
and reuse latency on whatever the build test environment is for
linux-ppc64-buildlet. I suspect this can happen on any linux
system, however, especially where the container has a limited PID
range.
The fix for this, limited to the test (the runtime syscall support
is unchanged), is to confirm that the Pid for the interrogated
thread's /proc/<TID>/status file confirms that it is still
associated with the test-process' PID.
linux-ppc64-buildlet:
go/bin/go test syscall -run=TestSetuidEtc -count=10000
ok syscall 104.285s
Fixes #42462
Change-Id: I55c84ab8361003570a405fa52ffec4949bf91113
Reviewed-on: https://go-review.googlesource.com/c/go/+/268717
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Tobias Klauser <tobias.klauser@gmail.com>
2020-11-09 20:28:04 -07:00
|
|
|
{call: "Setregid(101,0)", fn: func() error { return syscall.Setregid(101, 0) }, filter: "Gid:", expect: "\t101\t0\t0\t0"},
|
|
|
|
{call: "Setregid(0,102)", fn: func() error { return syscall.Setregid(0, 102) }, filter: "Gid:", expect: "\t0\t102\t102\t102"},
|
|
|
|
{call: "Setregid(0,0)", fn: func() error { return syscall.Setregid(0, 0) }, filter: "Gid:", expect: "\t0\t0\t0\t0"},
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
|
syscall: improve TestSetuidEtc() /proc/ parsing against races
TestSetuidEtc() was failing sporadically on linux-ppc64. From the
three https://build.golang.org/ logs, it looked like the logged
errors could be associated with threads dying, but proc reads
were, in some way, racing with their demise.
Exploring ways to increase thread demise, revealed that races
of this type can happen on non-ppc64 systems, and that
os.IsNotExist(err) was not a sufficient error condition test
for a thread's status file disappearing. This change includes a
fix for that to.
The actual issue on linux-ppc64 appears to be tied to PID reaping
and reuse latency on whatever the build test environment is for
linux-ppc64-buildlet. I suspect this can happen on any linux
system, however, especially where the container has a limited PID
range.
The fix for this, limited to the test (the runtime syscall support
is unchanged), is to confirm that the Pid for the interrogated
thread's /proc/<TID>/status file confirms that it is still
associated with the test-process' PID.
linux-ppc64-buildlet:
go/bin/go test syscall -run=TestSetuidEtc -count=10000
ok syscall 104.285s
Fixes #42462
Change-Id: I55c84ab8361003570a405fa52ffec4949bf91113
Reviewed-on: https://go-review.googlesource.com/c/go/+/268717
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Tobias Klauser <tobias.klauser@gmail.com>
2020-11-09 20:28:04 -07:00
|
|
|
{call: "Setreuid(1,0)", fn: func() error { return syscall.Setreuid(1, 0) }, filter: "Uid:", expect: "\t1\t0\t0\t0"},
|
|
|
|
{call: "Setreuid(0,2)", fn: func() error { return syscall.Setreuid(0, 2) }, filter: "Uid:", expect: "\t0\t2\t2\t2"},
|
|
|
|
{call: "Setreuid(0,0)", fn: func() error { return syscall.Setreuid(0, 0) }, filter: "Uid:", expect: "\t0\t0\t0\t0"},
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
|
syscall: improve TestSetuidEtc() /proc/ parsing against races
TestSetuidEtc() was failing sporadically on linux-ppc64. From the
three https://build.golang.org/ logs, it looked like the logged
errors could be associated with threads dying, but proc reads
were, in some way, racing with their demise.
Exploring ways to increase thread demise, revealed that races
of this type can happen on non-ppc64 systems, and that
os.IsNotExist(err) was not a sufficient error condition test
for a thread's status file disappearing. This change includes a
fix for that to.
The actual issue on linux-ppc64 appears to be tied to PID reaping
and reuse latency on whatever the build test environment is for
linux-ppc64-buildlet. I suspect this can happen on any linux
system, however, especially where the container has a limited PID
range.
The fix for this, limited to the test (the runtime syscall support
is unchanged), is to confirm that the Pid for the interrogated
thread's /proc/<TID>/status file confirms that it is still
associated with the test-process' PID.
linux-ppc64-buildlet:
go/bin/go test syscall -run=TestSetuidEtc -count=10000
ok syscall 104.285s
Fixes #42462
Change-Id: I55c84ab8361003570a405fa52ffec4949bf91113
Reviewed-on: https://go-review.googlesource.com/c/go/+/268717
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Tobias Klauser <tobias.klauser@gmail.com>
2020-11-09 20:28:04 -07:00
|
|
|
{call: "Setresgid(101,0,102)", fn: func() error { return syscall.Setresgid(101, 0, 102) }, filter: "Gid:", expect: "\t101\t0\t102\t0"},
|
|
|
|
{call: "Setresgid(0,102,101)", fn: func() error { return syscall.Setresgid(0, 102, 101) }, filter: "Gid:", expect: "\t0\t102\t101\t102"},
|
|
|
|
{call: "Setresgid(0,0,0)", fn: func() error { return syscall.Setresgid(0, 0, 0) }, filter: "Gid:", expect: "\t0\t0\t0\t0"},
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
|
syscall: improve TestSetuidEtc() /proc/ parsing against races
TestSetuidEtc() was failing sporadically on linux-ppc64. From the
three https://build.golang.org/ logs, it looked like the logged
errors could be associated with threads dying, but proc reads
were, in some way, racing with their demise.
Exploring ways to increase thread demise, revealed that races
of this type can happen on non-ppc64 systems, and that
os.IsNotExist(err) was not a sufficient error condition test
for a thread's status file disappearing. This change includes a
fix for that to.
The actual issue on linux-ppc64 appears to be tied to PID reaping
and reuse latency on whatever the build test environment is for
linux-ppc64-buildlet. I suspect this can happen on any linux
system, however, especially where the container has a limited PID
range.
The fix for this, limited to the test (the runtime syscall support
is unchanged), is to confirm that the Pid for the interrogated
thread's /proc/<TID>/status file confirms that it is still
associated with the test-process' PID.
linux-ppc64-buildlet:
go/bin/go test syscall -run=TestSetuidEtc -count=10000
ok syscall 104.285s
Fixes #42462
Change-Id: I55c84ab8361003570a405fa52ffec4949bf91113
Reviewed-on: https://go-review.googlesource.com/c/go/+/268717
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Tobias Klauser <tobias.klauser@gmail.com>
2020-11-09 20:28:04 -07:00
|
|
|
{call: "Setresuid(1,0,2)", fn: func() error { return syscall.Setresuid(1, 0, 2) }, filter: "Uid:", expect: "\t1\t0\t2\t0"},
|
|
|
|
{call: "Setresuid(0,2,1)", fn: func() error { return syscall.Setresuid(0, 2, 1) }, filter: "Uid:", expect: "\t0\t2\t1\t2"},
|
|
|
|
{call: "Setresuid(0,0,0)", fn: func() error { return syscall.Setresuid(0, 0, 0) }, filter: "Uid:", expect: "\t0\t0\t0\t0"},
|
syscall: support POSIX semantics for Linux syscalls
This change adds two new methods for invoking system calls
under Linux: syscall.AllThreadsSyscall() and
syscall.AllThreadsSyscall6().
These system call wrappers ensure that all OSThreads mirror
a common system call. The wrappers serialize execution of the
runtime to ensure no race conditions where any Go code observes
a non-atomic OS state change. As such, the syscalls have
higher runtime overhead than regular system calls, and only
need to be used where such thread (or 'm' in the parlance
of the runtime sources) consistency is required.
The new support is used to enable these functions under Linux:
syscall.Setegid(), syscall.Seteuid(), syscall.Setgroups(),
syscall.Setgid(), syscall.Setregid(), syscall.Setreuid(),
syscall.Setresgid(), syscall.Setresuid() and syscall.Setuid().
They work identically to their glibc counterparts.
Extensive discussion of the background issue addressed in this
patch can be found here:
https://github.com/golang/go/issues/1435
In the case where cgo is used, the C runtime can launch pthreads that
are not managed by the Go runtime. As such, the added
syscall.AllThreadsSyscall*() return ENOTSUP when cgo is enabled.
However, for the 9 syscall.Set*() functions listed above, when cgo is
active, these functions redirect to invoke their C.set*() equivalents
in glibc, which wraps the raw system calls with a nptl:setxid fixup
mechanism. This achieves POSIX semantics for these functions in the
combined Go and C runtime.
As a side note, the glibc/nptl:setxid support (2019-11-30) does not
extend to all security related system calls under Linux so using
native Go (CGO_ENABLED=0) and these AllThreadsSyscall*()s, where
needed, will yield more well defined/consistent behavior over all
threads of a Go program. That is, using the
syscall.AllThreadsSyscall*() wrappers for things like setting state
through SYS_PRCTL and SYS_CAPSET etc.
Fixes #1435
Change-Id: Ib1a3e16b9180f64223196a32fc0f9dce14d9105c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210639
Trust: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
2019-12-09 22:50:16 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
for i, v := range vs {
|
|
|
|
if err := v.fn(); err != nil {
|
|
|
|
t.Errorf("[%d] %q failed: %v", i, v.call, err)
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
if err := compareStatus(v.filter, v.expect); err != nil {
|
|
|
|
t.Errorf("[%d] %q comparison: %v", i, v.call, err)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|