flock.cc: Fix stack allocation from callee used in caller
* flock.cc (lockf_t::create_lock_obj_attr): Add buffer parameter.
Call _everyone_sd with buffer argument from caller rather than
everyone_sd with locally allocated stack buffer.
(lockf_t::create_lock_obj): Call create_lock_obj_attr only once
outside the loop and with additional buffer argument.
(lockf_t::open_lock_obj): Call create_lock_obj_attr with additional
buffer argument.
Corinna Vinschen [Sat, 29 Aug 2015 07:16:47 +0000 (09:16 +0200)]
Allow sysconf to return CPU cache information
* include/sys/unistd.h (_SC_LEVEL*): Add cache-related variables as
on Linux.
* fhandler_proc.cc (format_proc_cpuinfo): Fetch cache information
from new cache functions in sysconf.cc, get_cpu_cache_intel and
get_cpu_cache_amd.
* sysconf.cc (__nt_query_system): New local helper.
(get_nproc_values): Utilize __nt_query_system on pre-Windows 7 systems.
Use GetLogicalProcessorInformationEx otherwise to handle more than
64 CPUs. Only handle _SC_NPROCESSORS_CONF and _SC_NPROCESSORS_ONLN.
(get_phys_pages): New helper to handle _SC_PHYS_PAGES.
(cpuid2_cache_descriptor): New array to map Intel CPUID 2 descriptor
values to cache type, cache size, associativity and linesize.
(cpuid2_cache_desc_compar): Comparision function for bsearch over
cpuid2_cache_descriptor.
(get_cpu_cache_intel_cpuid2): New function to fetch cache info from
Intel CPUID 2.
(get_cpu_cache_intel_cpuid4): Ditto from Intel CPUID 4.
(get_cpu_cache_intel): New function as CPU-specific entry point.
(assoc): New array to map associativity values from AMD CPUID
0x80000006.
(get_cpu_cache_amd): New function to fetch cache info from AMD CPUIDs
0x80000005 and 0x80000006.
(get_cpu_cache): New function to fetch cache info.
(sca): Call get_phys_pages if _SC_PHYS_PAGES is requested. Call
get_cpu_cache for new _SC_* cache requests.
(SC_MAX): Set to _SC_LEVEL4_CACHE_LINESIZE.
(get_phys_pages(void)): Call get_phys_pages(int).
* include/cygwin/version.h (CYGWIN_VERSION_API_MINOR): Bump.
Corinna Vinschen [Thu, 27 Aug 2015 11:34:10 +0000 (13:34 +0200)]
Implement POSIX_MADV_WILLNEED/POSIX_MADV_DONTNEED for newer OSes
* autoload.cc (DiscardVirtualMemory): Import.
(PrefetchVirtualMemory): Import.
* mmap.cc (posix_madvise): Actually implement POSIX_MADV_WILLNEED
utilizing PrefetchVirtualMemory and POSIX_MADV_DONTNEED utilizing
DiscardVirtualMemory on systems supporting them.
* wincap.h (wincaps::has_broken_prefetchvm): New element.
* wincap.cc: Implement above element throughout.
(wincapc::init): Make sure has_broken_prefetchvm is only true on
W10 under WOW64.
* include/cygwin/version.h (CYGWIN_VERSION_DLL_MAJOR): Bump to 2003.
(CYGWIN_VERSION_API_MINOR): Reset to 0.
* new-features.xml (ov-new2.3): New section, document posix_madvise
POSIX_MADV_WILLNEED/POSIX_MADV_DONTNEED change.
Corinna Vinschen [Thu, 27 Aug 2015 10:55:22 +0000 (12:55 +0200)]
strlen-armv7.S: Fix preprocessor check
Hi!
I've got the situation, that the function strlen() occurs twice in libc.a
(building newlib for ARM-V7a and Size-Optimized).
In newlib/libc/machine/arm/strlen.c there are the pre-processor stetements ...
#if defined (__OPTIMIZE_SIZE__) || defined (PREFER_SIZE_OVER_SPEED) || \
(defined (__thumb__) && !defined (__thumb2__))
/*...*/
#else
and in newlib/libc/machine/arm/strlen-armv7.S the "exclude" begins with
/* NOTE: This ifdef MUST match the ones in arm/strlen.c
We fallback to the one in arm/strlen.c for size optimised or
for older architectures. */
#if defined(_ISA_ARM_7) || defined(__ARM_ARCH_6T2__) && \
!(defined (__OPTIMIZE_SIZE__) || defined (PREFER_SIZE_OVER_SPEED) || \
(defined (__thumb__) && !defined (__thumb2__)))
But this is not completely contrary to arm/strlen.c (see above)!
To fix the logical statement in arm/strlen-armv7.S there are parentheses needed
Corinna Vinschen [Tue, 25 Aug 2015 20:15:22 +0000 (22:15 +0200)]
autload.cc: Avoid clobbering return address in noload on i686
This fixes a long-standing problem when GetProcAddress fails
to load a function. The noload code calls SetLastError on
i686 without saving the edx register. Starting with Windows 7,
SetLastError apparently uses $edx and the register is set to
0x00000000 on return. So the subsequent `jmp *$edx' in noload
supposed to return to the caller, actually jumps to address NULL,
which results in a SEGV.
* autoload.cc (noload): i686 only: Save and restore $edx when calling
SetLastError to avoid clobbering return address stating with Windows 7.
James Greenhalgh [Tue, 25 Aug 2015 13:31:11 +0000 (14:31 +0100)]
Always declare "kill" in include/sys/signal.h
Hi,
As I mentioned recently [1], newlib is providing a "kill" symbol to link
against, without declaring "kill" in signal.h. This is confusing for the
libgfortran build, which tries to link against kill (which succeeds), then
tries to use it (which triggers -Werror=implicit-function-declaration).
This patch implements my suggestion in that thread - making the declaration
of 'kill' in libc/include/sys/signal.h unconditional.
I've tested this by building a modified libgfortran on AArch64/ARM to see
that the Werror goes away, and the libgfortran build succeeds.
Is something like this OK for newlib? If so, can someone please commit
it on my behalf, as I have no commit access here.
Otherwise, what is your preferred direction for me to take this patch?
Corinna Vinschen [Mon, 24 Aug 2015 16:37:53 +0000 (18:37 +0200)]
Fix hang stracing forking processes but not following child
* ntdll.h (PROCESSINFOCLASS): Define ProcessDebugFlags.
* sigproc.cc (child_info::child_info): Only propagate _CI_STRACED to
child if strace is actually tracing child processes.
Jon TURNEY [Tue, 11 Aug 2015 13:55:23 +0000 (14:55 +0100)]
Manuals failing to build on Ubuntu LTS 14.04
On 11/08/2015 11:08, Andre Vieira wrote:
> On 10/08/15 14:38, Jon TURNEY wrote:
>> On 07/08/2015 11:13, Andre Vieira wrote:
>>> Building the manuals on Ubuntu LTS 14.04 (64-bit) using pdfTeX
>>> 3.1415926-2.5-1.40.14 (TeX Live 2013/Debian) is failing with the
>>> following error:
>>> "Transcript written on libc.log.
>>> /usr/bin/texi2dvi: pdfetex exited with bad status, quitting."
>>>
>>> libc.log complaints about the following:
>>>
>>> You can't use `@unskip' in vertical mode.
>>>
[...]
>>>
>>> This code has been around for a while, so it might have to do with a
>>> change in pdfTex?
So it seems this problem has been around for a while, and looks like it
is due to a change in texi2dvi in texinfo 5.0 or late 4.x, see [1].
I guess it's not very noticeable since it only occurs when doing 'make
dvi', which doesn't happen by default.
Attached is a different and perhaps cleaner workaround to the one in
that thread.
From 4d386b5900b6c68e022004b447faa696be5ff8c7 Mon Sep 17 00:00:00 2001
From: Jon TURNEY <jon.turney@dronecode.org.uk>
Date: Tue, 11 Aug 2015 14:46:39 +0100
Subject: [PATCH] Use '@sp 1' rather than '@*' to workaround a change in
texi2dvi.
Since about TexInfo 5.0, using '@*' immediately after a table causes a 'You
can't use `@unskip' in vertical mode' error.
See https://sourceware.org/ml/newlib/2013/msg00057.html
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Andre Vieira [Wed, 5 Aug 2015 12:22:43 +0000 (13:22 +0100)]
Use machine header file for param configuration.
On 31/07/15 10:34, Richard Earnshaw wrote:
> On 31/07/15 10:28, Andre Vieira wrote:
>> newlib/ChangeLog:
>> 2015-07-28 Andre Vieira <andre.simoesdiasvieira@arm.com>
>>
>> * libc/sys/arm/sys/param.h: Include machine/param.h
>> (HZ, NOFILE, PATHSIZE): Define.
>>
>> param_refactor_1.patch
>>
>>
>> From abc2d5f3398721f6ca891b9581feaba58730b19c Mon Sep 17 00:00:00 2001
>> From: Andre Simoes Dias Vieira <andsim01@arm.com>
>> Date: Tue, 28 Jul 2015 12:10:59 +0100
>> Subject: [PATCH 1/2] Moved param configuration to machine/param.h
>>
>> ---
>> newlib/libc/sys/arm/sys/param.h | 12 ++++++++++--
>> 1 file changed, 10 insertions(+), 2 deletions(-)
>>
>> diff --git a/newlib/libc/sys/arm/sys/param.h b/newlib/libc/sys/arm/sys/param.h
>> index adc066e9a8756e07edaaa8cadc79b5f05c996ac9..622c371972ab3c9dbb93ea5c51323a593e2a171a 100644
>> --- a/newlib/libc/sys/arm/sys/param.h
>> +++ b/newlib/libc/sys/arm/sys/param.h
>> @@ -3,9 +3,17 @@
>> #ifndef _SYS_PARAM_H
>> # define _SYS_PARAM_H
>>
>> -# define HZ (100)
>> -# define NOFILE (60)
>> +#include <machine/param.h>
>> +
>> +#ifndef HZ
>> +# define HZ (60)
>
> Why have you changed the value for HZ? It seems that, by convention,
> ARM boards have always used 100.
>
> R.
>
>> +#endif
>> +#ifndef NOFILE
>> +# define NOFILE>(60)
>> +#endif
>> +#ifndef PATHSIZE
>> # define PATHSIZE (1024)
>> +#endif
>>
>> #define BIG_ENDIAN 4321
>> #define LITTLE_ENDIAN 1234
>>
>
Hi Richard,
ARM's machine/param.h that is included in "#include <machine/param.h>",
before the 'ifndef' already defines HZ to be 100. This file was already
there, it was just not being used. I understand that this 'ifndef' might
be confusing, though I decided to add it to mimic the behavior of the
default sys/param.h.
There is however an unrelated issue with this patch, a typo in the
"#define NOFILE" that crept in there due to some copy pasting when
splitting the patch.
Here is a fixed version.
BR,
Andre
newlib/ChangeLog:
2015-07-28 Andre Vieira <andre.simoesdiasvieira@arm.com>
* libc/sys/arm/sys/param.h: Include machine/param.h
(HZ, NOFILE, PATHSIZE): Define.
From abc2d5f3398721f6ca891b9581feaba58730b19c Mon Sep 17 00:00:00 2001
From: Andre Simoes Dias Vieira <andsim01@arm.com>
Date: Tue, 28 Jul 2015 12:10:59 +0100
Subject: [PATCH 1/2] Moved param configuration to machine/param.h
Corinna Vinschen [Mon, 17 Aug 2015 20:45:02 +0000 (22:45 +0200)]
Don't call LsaLookupSids if we're not utilizing Windows account DBs
* grp.cc (internal_getgrfull): Drop asking caches. Explain why.
(internal_getgroups): In case we're not utilizing the Windows account
DBs, don't call LsaLookupSids but iterate over the group SIDs in the
token and call internal_getgrsid for each of them. Explain why.
Corinna Vinschen [Mon, 17 Aug 2015 18:24:49 +0000 (20:24 +0200)]
Try harder to avoid LDAP access for RFC2307 mapping
* fhandler_disk_file.cc (fhandler_base::fstat_by_nfs_ea): Rearrange
to fall back to myself uid/gid in case we don't utilize Windows
account DBs, just as prior to 1.7.34.
* sec_helper.cc (cygpsid::get_id): Disable Samba user/group mapping per
RFC2307 if we're not utilizing Windows account DBs.
* security.cc (convert_samba_sd): Revert previous patch.
Corinna Vinschen [Sat, 15 Aug 2015 10:30:09 +0000 (12:30 +0200)]
Cygwin: Try to fix potential data corruption in pipe write
* fhandler.cc (fhandler_base_overlapped::raw_write): When performing
nonblocking I/O, copy user space data into own buffer. Add longish
comment to explain why.
* fhandler.h (fhandler_base_overlapped::atomic_write_buf): New member.
(fhandler_base_overlapped::fhandler_base_overlapped): Initialize
atomic_write_buf.
(fhandler_base_overlapped::fhandler_base_overlapped): New destructor,
free'ing atomic_write_buf.
(fhandler_base_overlapped::copyto): Set atomic_write_buf to NULL in
copied fhandler.
Corinna Vinschen [Fri, 14 Aug 2015 19:41:37 +0000 (21:41 +0200)]
Don't perform RFC2307 account mapping without account DB
* fhandler_disk_file.cc (fhandler_base::fstat_by_nfs_ea): Only try
to map user and group info per RFC2307 if account info is fetched
from Windows account DB.
(convert_samba_sd): Ditto.
Corinna Vinschen [Thu, 13 Aug 2015 15:59:47 +0000 (17:59 +0200)]
Fix /proc/cpuinfo topology and cache size info
* autoload.cc (GetLogicalProcessorInformationEx): Import.
(SetThreadGroupAffinity): Import.
* fhandler_proc.cc (add_size): New macro.
(get_msb): New inline function.
(mask_bits): Ditto.
(format_proc_cpuinfo): Drop handling of old CPUs. Check if we're
running on a OS version supporting porcessor groups. If so, use
SetThreadGroupAffinity to set thread affinity. Improve cache info
to include 3rd level cache on Intel CPUs. Improve multi core info.
* wincap.h (wincaps::has_processor_groups): New element.
* wincap.cc: Implement above element throughout.
Jeff Johnston [Fri, 7 Aug 2015 18:52:07 +0000 (14:52 -0400)]
or1k: Allow exception nesting
Allow exceptions to be nested, which is especially useful with urgent
interrupts while processing an exception.
The implementation counts up the nesting level with each call to an
exception. In the outer exception (level 1), the exception stack is
started. All nested exceptions just reserve the redzone (scratch
memory that may be used by compiler) and exception context on the
stack, but then process on the same scratch.
Restriction: Impure pointers are shared among all exceptions. This may
be solved by creating an impure data structure in the stack frame with
each nested exception.
Jon TURNEY [Wed, 29 Jul 2015 15:48:00 +0000 (16:48 +0100)]
Don't include stdio64 functions in reentrant syscalls menu when not bulding stdio64.
Fix documentation build since 6c2b1842 by not including stdio64 functions in the
reent syscalls menu if the node itself isn't going to be included because it's
under the STDIO64 flag.
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
This is an optimized memset for AArch64. Memset is split into 4 main
cases: small sets of up to 16 bytes, medium of 16..96 bytes which are
fully unrolled. Large memsets of more than 96 bytes align the
destination and use an unrolled loop processing 64 bytes per
iteration. Memsets of zero of more than 256 use the dc zva
instruction, and there are faster versions for the common ZVA sizes 64
or 128. STP of Q registers is used to reduce codesize without loss of
performance.
RtlFillMemory and RtlCopyMemory only work for size values
up to 2GB. Fix this problem by using NetBSD code for
memset and memcpy. Add entry points for memmove, wmemset,
wmemmove, wmemcpy. Thanks to Roman Petrovski
<RPetrovski@illumina.com> for pointing this out.
Sebastian Huber [Mon, 27 Jul 2015 09:31:26 +0000 (11:31 +0200)]
Move header files
During libgcc build the first include search path for <...> is
"../newlib/libc/sys/rtems/include". Move all RTEMS specific header
files to "libc/sys/rtems/include" so that they can be found. Later
during libc build the header files in the previous location were somehow
present, but for libgcc build they were invisible. This change is
necessary to use <pthread.h> for the GCC thread model implementation.
newlib/ChangeLog
2015-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
libc/sys/rtems/machine/_types.h: Move to ...
libc/sys/rtems/include/machine/_types.h: ... here.
libc/sys/rtems/machine/limits.h: Move to ...
libc/sys/rtems/include/machine/limits.h: ... here.
libc/sys/rtems/machine/param.h: Move to ...
libc/sys/rtems/include/machine/param.h: ... here.
libc/sys/rtems/sys/cpuset.h: Move to ...
libc/sys/rtems/include/sys/cpuset.h: ... here.
libc/sys/rtems/sys/dirent.h: Move to ...
libc/sys/rtems/include/sys/dirent.h: ... here.
libc/sys/rtems/sys/param.h: Move to ...
libc/sys/rtems/include/sys/param.h: ... here.
libc/sys/rtems/sys/syslimits.h: Move to ...
libc/sys/rtems/include/sys/syslimits.h: ... here.
libc/sys/rtems/sys/utime.h: Move to ...
libc/sys/rtems/include/sys/utime.h: ... here.
Sebastian Huber [Tue, 28 Jul 2015 09:32:05 +0000 (04:32 -0500)]
Add <sys/lock.h>
Provide self-contained synchronization objects for RTEMS. The API was
designed to be able to use it for the Newlib internal locks, the C11
threads support, the GCC thread model support and the libgomp operating
system configuration in combination with <pthread.h>.
cygwin: Fix crashes under AllocationPreference=0x100000 condition
* cygtls.h: Include cygtls_padsize.h and define CYGTLS_PADSIZE there.
* cygtls_padsize.h: New file. Define CYGTLS_PADSIZE.
* environ.cc (parse_options): Fix NULL pointer access.
* init.cc (threadfunc_fe): Do not force stack align on x86_64.
* strace.cc (main2): Rename from main.
(main): Make room for _cygtls area on stack and just call main2. Add
comment to explain why.
Jon TURNEY [Tue, 21 Jul 2015 12:41:46 +0000 (13:41 +0100)]
Use makedoc generated texinfo documentation for reentrant syscalls
We use texinfo nodes beginning with an underscore in several other places, so
revert this ancient workaround for a no longer existing bug, and use the makedoc
generated texinfo for reentrant versions of syscalls, rather than handwritten
documentation.
Also alphabetically sort these functions.
Also add documentation for _execve_r, _getpid_r, _kill_r and _times_r functions,
whose non-reentrant versions are documented as stubs
v2:
Keep _open64_r, _lseek64_r and _fstat64_r functions under texinfo conditional STDIO64
Add _stat64_r function likewise.
Notes:
1. The handwritten prototypes give the reentrancy structure pointer as of type
void *, rather than the presumably more correct struct __reent *
2. The fcntl, gettimeofday, mkdir and rename functions are not documented as
stubs, so I haven't added the reentrant versions either
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Fix potential buffer overflow in makecontext trampoline
glibc's tst-makecontext2 testcase uncovered a bug in
__cont_link_context. If the function misses to reserve
shadow space for the calls to setcontext/cygwin_exit,
both functions could overwrite memory beyond the stack
configured in uc_stack.
* exceptions.cc (__cont_link_context): x86_64: align stack and reserve
shadow space for subsequent function calls, otherwise suffer potential
buffer overflow.
* dcrt0.cc (initial_env): Reduce size of local path buffers to
PATH_MAX. Allocate debugger_command from process heap.
(init_windows_system_directory): Very early initialize new global
variable global_progname.
* dll_init.cc (dll_list::alloc): Make path buffer static. Explain why.
(dll_list::populate_deps): Use tmp_pathbuf for local path buffer.
* exceptions.cc (debugger_command): Convert to PWCHAR.
(error_start_init): Allocate debugger_command and fill with wide char
strings. Only allocate if NULL.
(try_to_debug): Just check if debugger_command is a NULL pointer to
return. Drop conversion from char to WCHAR and drop local variable
dbg_cmd.
* globals.cc (global_progname): New global variable to store Windows
application path.
* pinfo.cc (pinfo_basic::pinfo_basic): Just copy progname over from
global_progname.
(pinfo::status_exit): Let path_conv create the POSIX path to
avoid local buffer.
* pseudo_reloc.cc (__report_error): Utilize global_progname, drop local
buffer.
* smallprint.cc (__small_vsprintf): Just utilize global_progname for
%P format specifier.
(__small_vswprintf): Ditto.
* strace.cc (PROTECT): Change to reflect x being a pointer. Reformat.
(CHECK): Ditto. Reformat.
(strace::activate): Utilize global_progname, drop local buffer.
Fix formatting.
(strace::vsprntf): Reduce size of local progname buffer to NAME_MAX.
Copy and, if necessary, convert only the last path component to
progname.
(strace_buf_guard): New muto.
(buf): New static pointer.
(strace::vprntf): Use buf under strace_buf_guard lock only. Allocate
buffer space for buf on Windows heap.
* wow64.cc (wow64_respawn_process): Utilize global_progname, drop
local path buffer.
* exceptions.cc (__unwind_single_frame): Define empty macro on i686.
(_cygtls::call_signal_handler): Try to make sure signal context makes
sense in case we're generating context here. Add comment to explain.
* exceptions.cc (__unwind_single_frame): Move up in file to be
accessible from other places. Move comment to getcontext.
(stack_info::walk): Call __unwind_single_frame in 64 bit case. Fix
preceeding comment.
(myfault_altstack_handler): Call __unwind_single_frame.
(getcontext): Give comment from __unwind_single_frame a new home.
(swapcontext): Fix comment.
This is an optimized memset for AArch64. Memset is split into 4 main
cases: small sets of up to 16 bytes, medium of 16..96 bytes which are
fully unrolled. Large memsets of more than 96 bytes align the
destination and use an unrolled loop processing 64 bytes per
iteration. Memsets of zero of more than 256 use the dc zva
instruction, and there are faster versions for the common ZVA sizes 64
or 128. STP of Q registers is used to reduce codesize without loss of
performance.
This is an optimized memcpy for AArch64. Copies are split into 3 main
cases: small copies of up to 16 bytes, medium copies of 17..96 bytes
which are fully unrolled. Large copies of more than 96 bytes align
the destination and use an unrolled loop processing 64 bytes per
iteration. In order to share code with memmove, small and medium
copies read all data before writing, allowing any kind of overlap. On
a random copy test memcpy is 40.8% faster on A57 and 28.4% on A53.
This is an optimized memmove for AArch64. All copies of up to 96
bytes and all backward copies are done by the new memcpy. The only
remaining case is large forward copies which are done in the same way
as the memcpy loop, but copying from the end rather than the start.
x86_64: Handle myfault exceptions when running on alternate signal stack
x86_64 only:
* cygtls.cc (san::leave): Restore _my_tls.andreas.
* cygtls.h (class san): Add _clemente as in 32 bit case. Add ret and
frame members.
(san::san): Handle _my_tls.andreas as on 32 bit. Take parameter and
write it to new member ret. Store current stack pointer in frame.
(san::~san): New destructor to restore _my_tls.andreas.
(__try): Use __l_except address as parameter to san::san.
* dcrt0.cc (dll_crt0_0): Add myfault_altstack_handler as vectored
continuation handler.
* exception.h (myfault_altstack_handler): Declare.
* exceptions.cc (myfault_altstack_handler): New function. Explain what
it's good for.
Simplify stack allocation code in child after fork
* child_info.h (CURR_CHILD_INFO_MAGIC): Update.
(child_info_fork::alloc_stack_hard_way): Drop declaration.
* dcrt0.cc (child_info_fork::alloc_stack_hard_way): Fold into
child_info_fork::alloc_stack.
(getstack): Remove.
(child_info_fork::alloc_stack): Simplify check for application-provided
stack in "hard way" code. Don't call getstack for each page, just
reallocate stack immediately as required.
git://sourceware.org / newlib-cygwin.git / log