This CL adds TCPInfo struct to linux/386,arm.
It's already added to linux/amd64.
Note that not sure the reason but cgo godefs w/ latest gcc
translates a flexible array member in structures correctly,
handles it as a non-incomplete, non-opaque type, on Go 1.
This CL reverts such changes by hand for the Go 1 contract.
R=minux.ma, bradfitz, rsc
CC=golang-dev
https://golang.org/cl/7197046
Handle return values from recvfrom correctly when the
kernel decides to not return an address.
Fixes#4636.
Fixes#4352.
R=rsc, mikioh.mikioh, dave
CC=golang-dev
https://golang.org/cl/7058062
Fixes the fork-exec/wait race condition for ForkExec
as well, by making it use startProcess. This makes the
comment for StartProcess consistent as well.
Further, the passing of Waitmsg data in startProcess
and WaitProcess is protected against possible forks
from outside of ForkExec and StartProcess, which might
cause interference with the Await call.
R=rsc, rminnich, npe, ality
CC=golang-dev
https://golang.org/cl/7128059
Offsets for return values from seek were miscalculated
and a translation from 32-bit code for error handling
was incorrect.
R=rsc, rminnich, npe
CC=golang-dev
https://golang.org/cl/7181045
On Plan 9, only the parent of a given process can enter its wait
queue. When a Go program tries to fork-exec a child process
and subsequently waits for it to finish, the goroutines doing
these two tasks do not necessarily tie themselves to the same
(or any single) OS thread. In the case that the fork and the wait
system calls happen on different OS threads (say, due to a
goroutine being rescheduled somewhere along the way), the
wait() will either return an error or end up waiting for a
completely different child than was intended.
This change forces the fork and wait syscalls to happen in the
same goroutine and ties that goroutine to its OS thread until
the child exits. The PID of the child is recorded upon fork and
exit, and de-queued once the child's wait message has been read.
The Wait API, then, is translated into a synthetic implementation
that simply waits for the requested PID to show up in the queue
and then reads the associated stats.
R=rsc, rminnich, npe, mirtchovski, ality
CC=golang-dev
https://golang.org/cl/6545051
This decreases the amount of system calls during the
first call to Getenv. Calling Environ will still read
in all environment variables and populate the cache.
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/6939048
I've been writing some code which involves syncing files (like
rsync) and it became apparent that under Linux I could read
modification times (os.Lstat) with nanosecond precision but
only write them with microsecond precision. This difference
in precision is rather annoying when trying to discover
whether files need syncing or not!
I've patched syscall and os to increases the accuracy of of
os.Chtimes for Linux and Windows. This involved exposing the
utimensat system call under Linux and a bit of extra code
under Windows. I decided not to expose the "at" bit of the
system call as it is impossible to replicate under Windows, so
the patch adds syscall.Utimens() to all architectures along
with a ImplementsUtimens flag.
If the utimensat syscall isn't available (utimensat was added
to Linux in 2.6.22, Released, 8 July 2007) then it silently
falls back to the microsecond accuracy version it uses now.
The improved accuracy for Windows should be good for all
versions of Windows.
Unfortunately Darwin doesn't seem to have a utimensat system
call that I could find so I couldn't implement it there. The
BSDs do, but since they share their syscall implementation
with Darwin I couldn't figure out how to define a syscall for
*BSD and not Darwin. I've left this as a TODO in the code.
In the process I implemented the missing methods for Timespec
under Windows which I needed which just happened to round out
the Timespec API for all platforms!
------------------------------------------------------------
Test code: http://play.golang.org/p/1xnGuYOi4b
Linux Before (1000 ns precision)
$ ./utimetest.linux.before z
Setting mtime 1344937903123456789: 2012-08-14 10:51:43.123456789 +0100 BST
Reading mtime 1344937903123457000: 2012-08-14 10:51:43.123457 +0100 BST
Linux After (1 ns precision)
$ ./utimetest.linux.after z
Setting mtime 1344937903123456789: 2012-08-14 10:51:43.123456789 +0100 BST
Reading mtime 1344937903123456789: 2012-08-14 10:51:43.123456789 +0100 BST
Windows Before (1000 ns precision)
X:\>utimetest.windows.before.exe c:\Test.txt
Setting mtime 1344937903123456789: 2012-08-14 10:51:43.123456789 +0100 GMTDT
Reading mtime 1344937903123456000: 2012-08-14 10:51:43.123456 +0100 GMTDT
Windows After (100 ns precision)
X:\>utimetest.windows.after.exe c:\Test.txt
Setting mtime 1344937903123456789: 2012-08-14 10:51:43.123456789 +0100 GMTDT
Reading mtime 1344937903123456700: 2012-08-14 10:51:43.1234567 +0100 GMTDT
R=golang-dev, alex.brainman, rsc, bradfitz
CC=golang-dev
https://golang.org/cl/6905057
Thank you zhoumichaely for original CL 5175042.
Fixes#1740.
Fixes#2315.
R=golang-dev, bradfitz, mikioh.mikioh
CC=golang-dev, zhoumichaely
https://golang.org/cl/6822045
It is common to close network connection while another goroutine is
blocked reading on another goroutine. This sequence corresponds to
windows calls to WSARecv to start io, followed by GetQueuedCompletionStatus
that blocks until io completes, and, finally, closesocket called from
another thread. We were expecting that closesocket would unblock
GetQueuedCompletionStatus, and it does, but not always
(http://code.google.com/p/go/issues/detail?id=4170#c5). Also that sequence
results in connection is being reset.
This CL inserts CancelIo between GetQueuedCompletionStatus and closesocket,
and waits for both WSARecv and GetQueuedCompletionStatus to complete before
proceeding to closesocket. This seems to fix both connection resets and
issue 4170. It also makes windows code behave similar to unix version.
Unfortunately, CancelIo needs to be called on the same thread as WSARecv.
So we have to employ strategy we use for connections with deadlines to
every connection now. It means, there are 2 unavoidable thread switches
for every io. Some newer versions of windows have new CancelIoEx api that
doesn't have these drawbacks, and this CL uses this capability when available.
As time goes by, we should have less of CancelIo and more of CancelIoEx
systems. Computers with CancelIoEx are also not affected by issue 4195 anymore.
Fixes#3710Fixes#3746Fixes#4170
Partial fix for issue 4195
R=golang-dev, mikioh.mikioh, bradfitz, rsc
CC=golang-dev
https://golang.org/cl/6604072
Before this patch the test would close the file descriptor but
not the os.File. When the os.File was GC'ed, the finalizer
would close the file descriptor again. That would cause
problems if the same file descriptor were returned by a later
call to open in another test.
On my system:
> GOGC=30 go test
--- FAIL: TestPassFD (0.04 seconds)
passfd_test.go:62: FileConn: dup: bad file descriptor
FAIL
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/6776053
This is a part of a bigger change that adds data race detection feature:
https://golang.org/cl/6456044
The purpose of this patch is to provide coarse-grained synchronization
between all Read() and Write() calls.
R=rsc, bradfitz, alex.brainman
CC=golang-dev
https://golang.org/cl/6610064
This change updates CL 6576057 for exceptional cases where
return values from Syscall/RawSyscall functions are used.
The system calls return 32-bit integers. With the recent change
in size of `int' in Go for amd64, the type conversion was not
catching `-1' return values. This change makes the conversion
explicitly `int32'.
R=rsc, r
CC=golang-dev
https://golang.org/cl/6590047
The system calls return 32-bit integers. With the recent change
in size of `int' in Go for amd64, the type conversion was not
catching `-1' return values. This change makes the conversion
explicitly `int32'.
R=rsc, rminnich, npe, r
CC=golang-dev
https://golang.org/cl/6576057
This CL fixes code that incorrectly assumes that int is 32 bits wide.
Specifically, the socketpair system call expects a pointer to a pair
of int32s, not a pair of ints. Fix this inside the wrappers without
changing the APIs.
Update #2188.
R=golang-dev, r
CC=golang-dev
https://golang.org/cl/6552063
SO_PEERCRED on unix domain socket will fill a Ucred struct,
thus linux needs a custom Getsockopt variant.
Fixes#3836.
R=golang-dev, rsc, iant
CC=golang-dev
https://golang.org/cl/6445104
Fixes#3525.
PTRACE_SYSCALL behaves like PTRACE_CONT and can deliver
a signal to the process. Ideally PtraceSingleStep should
support the signal argument, but its interface is frozen
by Go1.
R=golang-dev, r, rsc
CC=golang-dev
https://golang.org/cl/6353051
This set of changes extends the Plan 9 support
to include the AMD64 architecture and should
work on all versions of Plan 9.
R=golang-dev, rminnich, noah.evans, rsc, minux.ma, npe
CC=akskuma, golang-dev, jfflore, noah.evans
https://golang.org/cl/6479052
Since NUL usually terminates strings in underlying syscalls, allowing
it when converting string arguments is a security risk, especially
when dealing with filenames. For example, a program might reason that
filename like "/root/..\x00/" is a subdirectory or "/root/" and allow
access to it, while underlying syscall will treat "\x00" as an end of
that string and the actual filename will be "/root/..", which might
be unexpected. Returning EINVAL when string arguments have NUL in
them makes sure this attack vector is unusable.
R=golang-dev, r, bradfitz, fullung, rsc, minux.ma
CC=golang-dev
https://golang.org/cl/6458050
This syscall was inadvertently exported when fixing
Getrlimit/Setrlimit on 32-bit platforms.
R=golang-dev, minux.ma
CC=golang-dev
https://golang.org/cl/6458051
Darwin kernels have a bug in accept() where error result from
an internal call is not checked and socket is accepted instead
of ECONNABORTED error. However, such sockets have no sockaddr,
which results in EAFNOSUPPORT error from anyToSockaddr, making
Go http servers running on Mac OS X easily susceptible to
denial of service from simple port scans with nmap.
Fixes#3849.
R=golang-dev, adg, mikioh.mikioh
CC=golang-dev
https://golang.org/cl/6456045
The previous version was not handling an off_t (64 bit)
parameter on 32 bit i386 systems. This patch splits sendfile
into two implementations in their respective arch-specific files.
Tested on FreeBSD amd64 and i386.
R=bradfitz
CC=golang-dev
https://golang.org/cl/6356048
This slipped in with the implementation of Getpid in CL 5909043.
I'd exclude that CL entirely but it is tangled up in the Win32finddata changes.
R=golang-dev, minux.ma
CC=golang-dev
https://golang.org/cl/6297065
Preserve old API by using correct struct in system call
and then copying the results, as we did for SetsockoptLinger.
R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/6307065
pipe2 is equivalent to pipe with flags set to 0.
However, pipe2 was only added recently. Using pipe
instead improves compatibility with NetBSD 5.
R=jsing
CC=golang-dev
https://golang.org/cl/6268045
RawSockaddrDatalink and SockaddrDatalink need to match - make Data
have length 12 for both.
R=golang-dev, mikioh.mikioh
CC=golang-dev
https://golang.org/cl/6223051
Fix mkerrors.sh so that it works on NetBSD.
Remove directory mode bits from types - this already appears in errors.
Regenerate the z* files now that cgo is working.
R=golang-dev, m4dh4tt3r, r
CC=golang-dev
https://golang.org/cl/6201077
Manually repair/update z* files for netbsd - this allows Go to
compile again on 386.
R=golang-dev, rsc, minux.ma
CC=golang-dev
https://golang.org/cl/6194064
Instead use a new type, "Note", whose underlying
type is just a string. This change allows us to
remove the exported os.Plan9Note type.
R=bradfitz, seed, rsc
CC=golang-dev
https://golang.org/cl/6015046
Now that Go will no longer work on OpenBSD versions prior to 5.1,
remove the sysctl workaround that was needed for 5.0 and earlier.
R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/6116050
1. Readdirnames was erroneously returning an
empty slice on every invocation.
2. The logic for determining which files to
close before exec was incorrect. If the
set of files to be kept open (provided by
the caller) did not include the files
opened at startup, those files would be
accidentally closed.
I also cleaned up readdupdevice while I was
in the vicinity.
R=golang-dev, seed, rsc
CC=golang-dev
https://golang.org/cl/6016044
syscall.Exit would originally kill only the calling
Go proc, leaving behind other procs in the
same group. This change makes syscall.Exit call
runtime·exit, which due to CL
https://golang.org/cl/5617048
will cleanly exit all the Go procs in the group.
R=golang-dev, rsc, rminnich, remyoudompheng, ality, john
CC=golang-dev, mirtchovski
https://golang.org/cl/6036051
This change comes from CL 5536043,
created by Andrey Mirtchovski. His
description follows:
"The plan9 exec child handler does not manage
dup-ed fds from the parent correctly: when a
dup-ed file descriptor appears in the child's fd
list it is closed when first encountered and then
subsequent attempt to dup it later in Pass 2 fails,
resulting in 'fork/exec: fd out of range or not
open'."
R=golang-dev, rminnich, ality
CC=golang-dev, mirtchovski, rsc
https://golang.org/cl/6009046
Building go currently fails on NetBSD, thus the regeneration
scripts do not run correctly. With these changes, at least
the go_bootstrap tool builds correctly.
R=golang-dev, r, bradfitz
CC=golang-dev
https://golang.org/cl/5695064
Tested using 6g and gccgo on x86_64 GNU/Linux and using gccgo
on PowerPC GNU/Linux (which is big-endian).
R=golang-dev, bradfitz, mikioh.mikioh, iant
CC=golang-dev
https://golang.org/cl/5975073
It is unprecedented to add tests to package syscall, especially
system-specific ones. Not a policy worth changing right before Go 1
is cut.
The sole existing test, passfd_test.go, contains the line
// +build linux darwin probablyfreebsd probablyopenbsd
which argues that this is not a subject to be undertaking likely.
Note that passfd_test.go also went in just now. It's the only test
in syscall.
Deleting for now, will reconsider after Go 1.
R=golang-dev, bradfitz, r, dsymonds
CC=golang-dev
https://golang.org/cl/5846063
This moves the various CA root fetchers from crypto/tls into crypto/x509.
The move was brought about by issue 2997. Windows doesn't ship with all
its root certificates, but will instead download them as-needed when using
CryptoAPI for certificate verification.
This CL changes crypto/x509 to verify a certificate using the system root
CAs when VerifyOptions.RootCAs == nil. On Windows, this verification is
now implemented using Windows's CryptoAPI. All other root fetchers are
unchanged, and still use Go's own verification code.
The CL also fixes the hostname matching logic in crypto/tls/tls.go, in
order to be able to test whether hostname mismatches are honored by the
Windows verification code.
The move to crypto/x509 also allows other packages to use the OS-provided
root certificates, instead of hiding them inside the crypto/tls package.
Fixes#2997.
R=agl, golang-dev, alex.brainman, rsc, mikkel
CC=golang-dev
https://golang.org/cl/5700087
CL 3075041 says ARM is not little-endian, but my test suggests otherwise.
My test program is:
package main
import ("fmt"; "syscall"; "os")
func main() {
err := syscall.Fallocate(1, 1/*FALLOC_FL_KEEP_SIZE*/, 0, int64(40960));
fmt.Fprintln(os.Stderr, err)
}
Without this CL, ./test > testfile will show: file too large; and strace shows:
fallocate(1, 01, 0, 175921860444160) = -1 EFBIG (File too large)
With this CL, ./test > testfile will show: <nil>; and strace shows:
fallocate(1, 01, 0, 40960) = 0
Quoting rsc:
"[It turns out that] ARM syscall ABI requires 64-bit arguments to use an
(even, odd) register pair, not an (odd, even) pair. Switching to "big-endian"
worked because it ended up using the high 32-bits (always zero in the tests
we had) as the padding word, because the 64-bit argument was the last one,
and because we fill in zeros for the rest of the system call arguments, up to
six. So it happened to work."
I updated mksyscall_linux.pl to accommodate the register pair ABI requirement,
and removed all hand-tweaked syscall routines in favor of the auto-generated
ones. These including: Ftruncate, Truncate, Pread and Pwrite.
Some recent Linux/ARM distributions do not bundle kernel asm headers,
so instead we always get latest asm/unistd.h from git.kernel.org (just like
what we do for FreeBSD).
R=ken, r, rsc, r, dave, iant
CC=golang-dev
https://golang.org/cl/5726051