arc-2016.09 is the most recent toolchain for ARC cores and
it is based on top of upstream Binutils 2.27 and GCC 6.2.1.
With updated major version of GCC we copied all GCC 6.x patches
for ARC as well as Bintils 2.27 patches.
Note that toochain sports ARCv4 ABI and so must be used
with 4.8+ Linux kernels. Even though it will build v4.4 kernel
perfectly fine on attempt to run user-space apps they won't
work with older kernel. That said previuosly sent RFC patches with
Linux kernel update are required:
[1] https://patchwork.ozlabs.org/patch/726686/
[2] https://patchwork.ozlabs.org/patch/726687/
Signed-off-by: Alexey Brodkin <Alexey.Brodkin@synopsys.com>
Binutils recently updated the ABI version to 5, which broke glibc, as it
refuses to load anything > 3. Upstream reverted the ABI version update,
so backport the commit.
Fixes the following error on boot:
/bin/sh: error while loading shared libraries: /lib/libcrypt.so.1: ELF file ABI version invalid
[ 0.920077] Kernel panic - not syncing: Attempted to kill init! exitcode=0x00007f00
[ 0.920077]
[ 0.929450] Rebooting in 1 seconds..
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
- Check for changes in toolchain/ only
- Only replace the stamp file if it was changed
- Fix dependencies for staging dir prepare
- Move the stamp file to TOOLCHAIN_DIR
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Important change was made in 1.0.18: all sub-libs were merged
in one and only libc similarly to musl.
See [1] for more details.
To support that we had to remove refences to those sub-libs like
libpthread, libcrypt, libdl, libm, libutil etc.
[1] http://cgit.uclibc-ng.org/cgi/cgit/uclibc-ng.git/commit/?id=29ff9055c80efe77a7130767a9fcb3ab8c67e8ce
Signed-off-by: Alexey Brodkin <Alexey.Brodkin@synopsys.com>
Currently on attempt to configure uClibc with "V=s" user is prompted
to answer on configuration questions. Major inconvenience here is
build procedure get paused until user answers all questions.
That happens because not all symbols are defined in our .config.
Solution is as simple as usage of "olddefconfig" target instead of
simple "oldconfig". In that case default values get silently set
for symbols not defined by our .config explicitly.
We haven't noticed that subtle issue before because without "V=s"
what we do is "make oldconfig < /dev/null" which automatically
answers all questions :)
Signed-off-by: Alexey Brodkin <Alexey.Brodkin@synopsys.com>
If the staging dir was deleted, the build needs to recompile some files.
This change speeds up this corner case significantly
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Use 2.27 as default which is the current relese.
Makes it easier to get support and submit patches upstream if needed.
Tested on ar71xx, mt7621, kirkwood (arm)
Signed-off-by: Daniel Engberg <daniel.engberg.lists@pyret.net>
This reverts commit dde5c729ec.
The uClibc update was completely untested, does not build with Kernel 4.4 and
did not adjust the configuration to predeclare new config symbols.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
This patch updates musl to 1.1.16+ [0] and removes all the
backported patches. This is a major release and tagged as such.
For more information visit musl-libc.org or read the WHATSNEW.
Furthermore, this patch also changes musl to download directly
from git. This makes it easier to update musl in the future.
The patch custom Add-format-attribute-to-some-function-declarations.patch
was assigned a new 400- number. This should avoid confusion
since 0xx numbers are usually assigned to backports.
[0] <http://git.musl-libc.org/cgit/musl/commit/?id=769f53598e781ffc89191520f3f8a93cb58db91f>
Cc: Hannu Nyman <hannu.nyman@iki.fi>
Cc: Koen Vandeputte <koen.vandeputte@ncentric.com>
Cc: Jo-Philipp Wich <jo@mein.io>
Cc: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
Backport most important fixes up to latest HEAD
- Taken post-commit reverts/fixes into account
Compile tested
Run-tested on cns3xxx & imx6 targets
Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
graphite is disabled by default and it's not worth carrying ppl and
cloog for supporting this feature on the old gcc 4.8, which is only used
for ARC
Signed-off-by: Felix Fietkau <nbd@nbd.name>
This is intended to be used for a wide array of package sanity checks.
The first check that is implemented is for the hash of downloaded files.
It checks:
- Missing hash
- Use of SHA256 instead of MD5
- dl/<file> hash not matching hash in makefile
- deprecated MD5SUM variable
The deprecated MD5SUM variable check is skipped for feeds/ until OpenWrt
is updated as well
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Otherwise the built binaries may fail with
Error relocating /usr/bin/qemu-system-x86_64: unsupported relocation type 37
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
fix parsing of the < > quoted time zone names. Compare the correct
character instead of repeatedly comparing the first character.
Patch has been submitted to upstream as
http://www.openwall.com/lists/musl/2016/10/24/3
Longer explanation in
http://www.openwall.com/lists/musl/2016/10/19/1
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
[Jo-Philipp Wich: move references into commit message, shorten title prefix]
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
Add binutils 2.27 to toolchain
Signed-off-by: Daniel Engberg <daniel.engberg.lists@pyret.net>
Signed-off-by: Felix Fietkau <nbd@nbd.name> [refresh patches]
Make it a choice menu which offers the 3 C libraries we know about: glibc,
uClibc and musl. While at it, make it possible for the external toolchain libc
to select USE_GLIBC, USE_UCLIBC or USE_MUSL which is used by several packages
to conditionally include specific CFLAGS (e.g: iproute2).
Because USE_GLIBC et al. can now be selected by external toolchains, we need to
restrict the per-libc menus to check on !EXTERNAL_TOOLCHAIN.
While at it, make musl the default C library for external toolchain to match
the internal toolchain.
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Configurations without shadow passwords have been broken since the removal
of telnet: as the default entry in /etc/passwd is not empty (but rather
unset), there will be no way to log onto such a system by default. As
disabling shadow passwords is not useful anyways, remove this configuration
option.
The config symbol is kept (for a while), as packages from feeds depend on
it.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
since eglibc is deprecated just use glibc as name
- fix build errors w. stack-protector (observed for x86_64):
libresolv_pic.a(gethnamaddr.os): In function `addrsort':
gethnamaddr.c:(.text+0x425): undefined reference to `__stack_chk_guard'
by additionally setting libc_cv_ssp_strong=no
like previously libc_cv_ssp=no was set
- fix compile errors on 2.24 x86 introduced by
upstream changesets "Optimize i386 syscall inlining"
errors are like:
- https://sourceware.org/ml/libc-alpha/2015-10/msg00745.html
and trying to fix with the whats suggested at:
https://patchwork.openembedded.org/patch/118909/
leads to other error:
gcc6: elf/librtld.os: In function `__mmap':
(.text+0x131a9): undefined reference to `__libc_do_syscall'
or:
gcc5: elf/dl-load.os
{standard input}: Assembler messages:
{standard input}:5129: Error: symbol `__x86.get_pc_thunk.cx' is already defined
instead of testing other flags/effects (-fno-omit-frame-pointer)
just use -O2 (like buildroot does) instead of -Os
boot+pings tested on:
qemu malta (le+be) (gcc5+bin2.25.1)
qemu aarch64 (gcc5+bin2.25.1)
qemu x86_64 (gcc6+bin2.26)
qemu x86 generic (gcc6+bin2.26)
Signed-off-by: Dirk Neukirchen <dirkneukirchen@web.de>
-fno-plt and musl-libc patches already get upstream, then dropped
in GCC 6. Other patches are almost identical compared to GCC 5.4.
Compile and run tested on ar71xx/Qihoo C301 and
mvebu/Linksys WRT1900ac v1
Signed-off-by: Syrone Wong <wong.syrone@gmail.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name> [cleanups/fixes]
Use ./patches and ./patches-arc instead of version dependent patches.
We should have only one main supported version (along with the special
case for arc).
Signed-off-by: Felix Fietkau <nbd@nbd.name>
The removed patch is already integrated in upstream musl.
Add one additional patch which fixes a regression on mips.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
No extra libc header build step is done, so no extra toolchain is needed
for preparing it.
This saves a significant amount of build time and disk space
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Even though ARC tools were released as arc-2016.03
we have a little bit different version name for GDB.
That's because both Binutils and GDB come from the same
git repo but from different branches (and so different tags).
Also removing an extra patch that made its way into release.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
This change switches ARC tools to the most recent arc-2016.03
version.
ARC GNU tools of version arc-2016.03 bring some quite significant
changes like:
* Binutils v2.26+ (upstream commit id 202ac19 with additional ARC
* patches)
* GCC v4.8.5
* GDB 7.10
More about changes, improvements and fixes could be found here:
https://github.com/foss-for-synopsys-dwc-arc-processors/toolchain/releases/tag/arc-2016.03
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
This is required to build net-snmp.
If that options is disabled in uClibc then net-snmp doesn't
detect sighold support on its configuration stage and so
defines HAVE_SIGHOLD as 0. That in its turn causes compilation of
the following branch of timerPause() in apps/snmpnetstat:
------------------>8----------------
int oldmask;
oldmask = sigblock(sigmask(SIGALRM));
if (!signalled) {
sigpause(0);
}
sigsetmask(oldmask);
------------------>8----------------
Now in uClibc all 3 sigblock(), sigmask() and sigsetmask()
were removed back in 2005, see:
https://git.busybox.net/uClibc/commit/?id=5aa7aa7fa7ec2a0fe567ac0b2595b46add6f3594
And all that causes net-snmp linkage to fail this way:
------------------>8----------------
.libs/if.o: In function `intpr':
if.c:(.text+0x908): undefined reference to `sigmask'
if.c:(.text+0x90e): undefined reference to `sigblock'
if.c:(.text+0x924): undefined reference to `sigsetmask'
------------------>8----------------
If we enable UCLIBC_HAS_OBSOLETE_BSD_SIGNAL in uClibc then
branch with sighold is used in timerPause() and everything builds as
expected.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 49178
This change adds support of ARC ISAv2 processors in
OpenWRT toolchain.
In general gcc for ARC may compile code for both ISA versions
simultaneously but libgcc will be built only for default
architecture that's why it's necessary to specify --with-cpu
on gcc configuration.
As for uClibc we need to use different configurations for
different ARC ISAs.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Felix Fietkau <nbd@openwrt.org>
Cc: Jo-Philipp Wich <jow@openwrt.org>
Cc: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 48739
Refresh patches and rework Makefile to fetch glibc from release branches
instead of relying on tarballs.
Signed-off-by: Jo-Philipp Wich <jow@openwrt.org>
SVN-Revision: 48735
mips returns -ENOSYS in case it can not handle the vdso call and wants
the libc to call the original syscall in such a case. This fixes the
patch to add such handling. I hope this fixes the random reboots I got.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 48501
GCC and Clang are able to check the format arguments given to a
function and warn the user if there is a error in the format arguments
or if there is a potential uncontrolled format string security problem
in the code. GCC does this automatically for some functions like
printf(), but it is also possible to annotate other functions in a way
that it will check them too. This feature is used by glibc for many
functions. This patch adds the attribute to the some functions of musl
expect for these functions where gcc automatically adds it.
GCC automatically adds checks for these functions: printf, fprintf,
sprintf, scanf, fscanf, sscanf, strftime, vprintf, vfprintf and
vsprintf.
The documentation from gcc is here:
https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html
The documentation from Clang is here:
http://clang.llvm.org/docs/AttributeReference.html#format-gnu-format
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 48467
vdso support is available on mips starting with kernel 4.4, see kernel
commit a7f4df4e21 "MIPS: VDSO: Add implementations of gettimeofday()
and clock_gettime()" for details.
These are my micro benchmark results for 1.000.000 calls to
clock_gettime(CLOCK_MONOTONIC, &tp)
without vdso:
root@OpenWrt:/# time ./vdso-test
real 0m 0.95s
user 0m 0.24s
sys 0m 0.70s
with vdso:
root@OpenWrt:/# time /usr/bin/vdso-test
real 0m 0.35s
user 0m 0.34s
sys 0m 0.00s
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
SVN-Revision: 48466
This will make adding future glibc versions easier because the
conditionals won't have to be modified again.
Signed-off-by: Michael Marley <michael@michaelmarley.com>
SVN-Revision: 48399
Fixes: commit f17e56eff5b9 ("gcc: remove version 4.6, it is no longer needed")
CC: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Sedat Dilek <sedat.dilek@gmail.com>
SVN-Revision: 47965
