Regression testing newlib in conjunction with libgloss and
--enable-multilib can result in incompatible multilib versions of
newlib and libgloss being used during link.
This manifests on ARM target when newlib regression is run using a
GCC configured using --with-multilib-list=aprofile
With this configuration many of the multilib variants built are
mutually incompatible.
The issue is that the newlib dejagnu foo iterates each multilib
variant and correctly chooses the appropriate newlib variant but
always chooses the root/base libgloss variant.
The implementation of newlib/testsuite/lib/flags.exp contains the
following fragment:
set target_build_path "$objdir/$multibuildtop.."
The effect of this fragment is to explicitly select the root version
of libgloss, irrespective of the current multilib.
Digging around in VC it appears that the original implementation of
multlib magic came into the tree back in 2002 with:
6e9d950a (Thomas Fitzsimmons 2002-05-01 17:06:25 +0000 39)
In this initial version of multilib support, newlib was multilib
capable, but libgloss was not multilib capable, hence the necessity to
explicitly select the root libgloss version.
Subsequently flags.exp was modified to support out of tree testing:
cec1d3b4 (Jeff Johnston 2005-07-05 00:11:50 +0000 25)
This change is orthogonal to this issue, its effect is to exit early
in none multilib configurations.
Subsequently libgloss gained --enable-multilib support, the relevant
change is:
https://sourceware.org/ml/newlib/2006/msg00440.html
commit 00a4b31ad0
Author: Jeff Johnston <jjohnstn@redhat.com>
Date: Wed May 10 20:51:41 2006 +0000
This change enabled multilib support throughout libgloss, but ommitted
to adjust the flag.exp behaviour which anchors the libgloss multilib
selection to the base version.
The attached patch adjusts flags.exp to select the current multilib
variant of libgloss.
2015-11-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
* testsuite/lib/flags.exp (libgloss_link_flags): Drop multilibtop
from target_build_path.
Align makedoc's iscommand() with it's documentation, and don't allow commands to
contain a space. A command is a line containing only a sequence of capital
letters or '_', followed by optional spaces.
This prevents "IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
PARTICULAR PURPOSE" (the only line in the license text which consists of only
capitals and spaces, without any punctuation) from being interepreted as a
makedoc command, leading to:
"Can't find IMPLIED"
"warning, IMPLIED is not recognised"
being emitted by makedoc, (which is normally un-noticed because makedoc's stderr
is redirected to a .ref file)
2015-11-06 Jon Turney <jon.turney@dronecode.org.uk>
* doc/makedoc.c (iscommand): Only allow commands to have trailing
spaces, not space separated words.
Signed-off-by: Jon Turney <jon.turney@dronecode.org.uk>
makedoc defines a command as 'all upper case, and alone on a line'.
A few QUICKREF lines currently violate this by having some additional text after
the QUICKREF.
So, currently, these lines are treated as an unknown command.
This is benign as QUICKREF currently does nothing but produce some ignored
output on stderr. I'm not sure what the intent of QUICKREF is.
2015-11-06 Jon Turney <jon.turney@dronecode.org.uk>
* libm/mathfp/s_acos.c: Fix QUICKREF.
* libm/mathfp/e_acosh.c: Ditto.
* libm/math/w_asin.c: Ditto.
* libm/mathfp/e_acosh.c: Ditto.
* libm/mathfp/s_acos.c: Ditto.
Signed-off-by: Jon Turney <jon.turney@dronecode.org.uk>
The unused INTERNAL_DEFINITION, INTERNAL_FUNCTION and INTERNAL commands are
defined in terms of the non-existent built-in 'func'
This causes every single invocation of makedoc to output "Can't find func" three
times, as it parses doc.str.
This is normally un-noticed because makedoc's stderr is redirected to a .ref
file.
Fix these unused command definitions to something with equivalent lack of
effect, but without generating an error.
2015-11-06 Jon Turney <jon.turney@dronecode.org.uk>
* doc/doc.str: Fix INTERNAL_DEFINITION, INTERNAL_FUNCTION and
INTERNAL.
Signed-off-by: Jon Turney <jon.turney@dronecode.org.uk>
ARM newlib has various strcmp implementations that use .cfi_*
directives to generate unwind information.
The effect of this is that the generated objects contain .eh_frame
sections. However, ARM uses its own unwind info format, not
.eh_frame, which is generated by ARM-specific directives, not .cfi_*.
The .eh_frame sections are useless, but also not removed by strip and
may be loaded into memory at runtime.
This patch fixes this by using .cfi_sections .debug_frame (as in
glibc) so that the directives generate .debug_frame instead.
.debug_frame is useful for the debugger, can be removed by strip, and
is not loaded into memory at runtime.
* libc/machine/arm/strcmp-arm-tiny.S: Use .cfi_sections
.debug_frame.
* libc/machine/arm/strcmp-armv4.S: Likewise.
* libc/machine/arm/strcmp-armv4t.S: Likewise.
* libc/machine/arm/strcmp-armv6.S: Likewise.
* libc/machine/arm/strcmp-armv6m.S: Likewise.
* libc/machine/arm/strcmp-armv7.S: Likewise.
* libc/machine/arm/strcmp-armv7m.S: Likewise.
The patch cleans up the auto configury mechanism used to select
different implementations of memchr for various architecture versions.
The approach here is to remove the selection of memchr within automake
and instead use complimentary logic in memchr-stub.c and memchr.S to
choose between the gerneric memchr.c implementation or one of the
architecture specific implementations.
This patch also changes the selection criteria inline with the
previous proposal here:
https://sourceware.org/ml/newlib/2015/msg00752.html
but using the ACLE predefines.
Regressed for armv7-a armv5 armv8-a, correct selection of memcpy
implementation by manual inspection of a test program built for these
three architectures.
This patch cleans up the auto configury mechanism used to select
different implementations of memcpy for various architecture versions.
The approach here is to remove the selection of memcpy within automake
and instead use complimentary logic in memcpy-stub.c and memcpy.S to
choose between the generic memcpy.c implemenation or one of the
architecture specific memcpy*.S implemenations.
Regressed for armv7-a armv5 armv8-a, correct selection of memcpy
implementation by manual inspection of a test program built for these
three architectures.
This revised patch flips the remaining preprocessor logic in
memcpy-stub.c to use ACLE defines as requested in the previous review
and removes the now disused HAVE_ARMV7A and HAVE_ARMV8A configure.in
support.
The newlib configury logic that detects architecture version and
chooses an appropriate memcpy implementation does not consider
ARMv8-a.
This patch adds configury logic to detect ARMv8-a along with the
associated changes in Makefile.am and memcpy.
This is mandated by POSIX.
2015-10-30 Sebastian Huber <sebastian.huber@embedded-brains.de>
libc/include/pthread.h: Include <sched.h> instead of
<sys/sched.h>.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
aligned_alloc() is implemented in terms of posix_memalign() which is
only declared in <stdlib.h> but not defined in Newlib in general. At
least Linux and RTEMS implement this function.
newlib/ChangeLog
2015-10-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
* libc/stdlib/Makefile.am (GENERAL_SOURCES): Add
alloc_aligned.c.c.
* libc/stdlib/Makefile.in: Regenerate.
* libc/stdlib/aligned_alloc.c: New.
Import some <stdlib.h> function declarations from latest FreeBSD and
implement them. I am not sure if we should call the global reent
cleanup in quick_exit() similar to exit().
newlib/ChangeLog
2015-10-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
* libc/include/stdlib.h (at_quick_exit): Declare.
(quick_exit): Likewise.
* libc/stdlib/Makefile.am (GENERAL_SOURCES): Add
quick_exit.c.
* libc/stdlib/Makefile.in: Regenerate.
* libc/stdlib/quick_exit.c: New.
- Move types and defines to
<machine/_threads.h> so that it can be customized per target.
* libc/include/threads.h: New.
* libc/sys/rtems/include/machine/_threads.h: Likewise.
* 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.
* new-features.xml (ov-new2.3): Document sysconf cache addition.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
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
#if !(defined(_ISA_ARM_7) || defined(__ARM_ARCH_6T2__))
/*...*/
#endif
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
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
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?
Thanks,
James
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.
[1] https://sourceware.org/ml/newlib/2013/msg00057.html
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>
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
* libc/include/sys/signal.h [__CYGWIN__]: include <sys/ucontext.h>
if compiling for POSIX.1-2008.
Signed-off-by: Yaakov Selkowitz <yselkowi@redhat.com>
Only on first call to the recursive malloc lock the restore value of
exception enable fields is stored.
* libc/sys/or1k/mlock.c: Fix exception enable saving
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>
Sorry, there was a typo in <sys/lock.h> which leads to memory corruption
since not enough space is reserved for the lock object.
newlib/ChangeLog
2015-07-30 Sebastian Huber <sebastian.huber@embedded-brains.de>
* libc/sys/rtems/include/sys/lock.h (__LOCK_INIT_RECURSIVE): Use
proper type.
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.
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.
This got removed everywhere else in commit 139f923b, so I'm assuming whatever
ancient bug this was has been fixed long since.
There are plenty of other uses of texinfo with a filename containing an
underscore now.
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Anton Kolesov