| Age | Commit message (Collapse) | Author |
|---|
|
License notices in most of the source files refer to an outdated
FSF address. Replace it with URL, like in the rest of the source
files.Signed-off-by: Nikola Forró <nforro@redhat.com>
|
|
Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
|
|
On systems where uClibc doesn't provide an arch specific byteswap.h,
we fall back on bits/byteswap-common.h. However, there is a bug
in this header in the __bswap_constant_64(x) macro. If, for example,
a double is passed, we get 'invalid operands to binary &' in which we
mismatch a 'double' and 'long long unsigned int'. The newer glibc
headers fix this and so we import them. Also, since the inclusion of
byteswap.h is not appropriate for assembly code, we guard a portion
of endian.h which uses byteswap.h from inclusion in any assembly.
This is needed, for example, for f2fs-tools 1.6.0 on 32-bit big
endian PowerPC.
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
Currently, the Thumb support on ARM has three related Config.in
options, which are not trivial for users to understand, and are in
fact not needed:
- The USE_BX option is not needed: knowing whether BX is available or
not is easy. If you have an ARM > v4 or ARMv4T, then BX is
available, otherwise it's not. This is the logic used in glibc.
- The USE_LDREXSTREX option is not needed: whenever Thumb2 is
available, ldrex/strex are available, so we can simply rely on
__thumb2__ to determine whether ldrex/strex should be used, without
requiring a Config.in option.
- Once USE_BX and USE_LDREXSTREX are removed, the only thing left
that COMPILE_IN_THUMB does is to set -mthumb. This makes the option
unnecessary, as on ARM at least, the user is already supposed to
pass -march=<foo> or other compiler options tuning the library for
a specific ARM variant. There is no reason to do otherwise for
Thumb, which allows to get rid of the COMPILE_IN_THUMB option.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
|
|
Register window saving loop in __sigsetjmp incorrectly calculates offset
in the jmp_buf in case it was called with call12. The bug was introduced
in commit f4b1b7ade7971141 "xtensa: fix setjmp that didn't save all
registers correctly".
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
|
|
|
|
__JMP_BUF_SP is the index of a stack pointer slot in the __jmp_buf.
According to ARM __sigsetjmp code it does not depend on configuration
parameter __UCLIBC_HAS_FPU__. Make its definition unconditional.
__JMP_BUF_SP is used in the unwind_stop through the
_JMPBUF_CFA_UNWINDS_ADJ macro, this change fixes cleanup routines call
on thread cancellation in configurations with hard FP.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
|
|
For some xtensa cores generated code gets arranged so that
__default_sa_restorer immediately follows preceding function. E.g.:
40a1b: c03340 sub a3, a3, a4
40a1e: f01d retw.n
00040a20 <__default_sa_restorer>:
40a20: e1a022 movi a2, 225
40a23: 005000 syscall
...
Contents of the .eh_frame section:
...
000007f4 00000014 000007f8 FDE cie=00000000 pc=0004076c..00040a20
DW_CFA_advance_loc4: 3 to 0004076f
DW_CFA_def_cfa_offset: 48
Due to the way xtensa libgcc uw_frame_state_for calls _Unwind_Find_FDE
for non-signal frames, FDE for the previous function is found during
thread cancellation signal stack unwinding in that case. Signal stack
frame is not recognized and is not unwound properly, breaking cleanup
routines calling for cancelled thread.
Insert padding before the __default_sa_restorer so that no FDE is found
for it, MD_FALLBACK_FRAME_STATE_FOR is called by uw_frame_state_for and
the frame is correctly recognized as signal frame.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
|
|
The FR-V port is really broken, and I have no emulator
or hardware for this platform. I tried to get some hardware
from RedHat, who made the FR-V port initially. Unfortunately
Fujitsi didn't agreed to sent me some of their unused spare
hardware lying @RedHat. As I invested some time to get stuff compiled,
I decided to add the code and may be anytime later I can gain
access to some emulator or hardware.
GDB simulator for FR-V doesn't support booting Linux AFAIK.
|
|
ISC bind9 uses ptrsize, better use something else.
Reported-By: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Waldemar Brodkorb <wbx@uclibc-ng.org>
|
|
Althought recent versions of sofware don't require it, old ntpd
wants - see http://bugs.ntp.org/show_bug.cgi?id=769
Signed-off-by: Leonid Lisovskiy <lly.dev@gmail.com>
Signed-off-by: Waldemar Brodkorb <wbx@uclibc-ng.org>
|
|
Some software make use of exp10, so add it.
Basic testing with LTP.
Signed-off-by: Waldemar Brodkorb <wbx@uclibc-ng.org>
|
|
At least allow to compile a toolchain targeting nios2 without MMU.
|
|
At least allow to build a toolchain for hppa.
Sync some headers with glibc.
|
|
libc/sysdeps/linux/common/euidaccess.c:1:1: warning: "_GNU_SOURCE" redefined
In file included from <command-line>:0:
./include/libc-symbols.h:52:1: warning: this is the location of the previous definition
Signed-Off-By: Leonid Lisovskiy <lly.dev@gmail.com>
|
|
Different define of latest kernel.
kernel define is
#define O_DIRECT 00040000 /* direct disk access hint */
#define O_DIRECTORY 00200000 /* must be a directory */
Yoshinori Sato <ysato@users.sourceforge.jp>
|
|
Patch from OpenWrt.
Reported-by: Leonid Lisovskiy <lly.dev@gmail.com>
|
|
Reported-by: Leonid Lisovskiy <lly.dev@gmail.com>
|
|
open_cancel and open2_cancel have diffrent argument on open.
So can't alias this functions.
Signed-off-by: Yoshinori Sato <ysato@users.sourceforge.jp>
|
|
If argument passing to register case (ex. -mregparam=3).
This case set all parameters set to register from caller.
But callee refer to stack. So can't get parameter.
Signed-off-by: Yoshinori Sato <ysato@users.sourceforge.jp>
|
|
Signed-off-by: Yoshinori Sato <ysato@users.sourceforge.jp>
|
|
Signed-off-by: Yoshinori Sato <ysato@users.sourceforge.jp>
|
|
syncfs() was recently added (commit dfa593d4d). But man sync(2) specifies
that syncfs() is Linux-specific. This was overlooked in the original
commit so we add it to the set of Linux-specific functions supported by
uClibc.
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
One test fails to compile with a gcc ICE, so disable it for now.
|
|
|
|
Both architectures are more or less deprecated.
No Linux upstream support, no gcc support for uClinux.
|
|
Argp is an advanced support for parsing unix-style argument vectors.
In addition to the common getopt interface, it provides automatic
response
to `--help' and `--version' options and use of custom parser in
conjunction
with argp native option parser, among others.
Argp support is required by elfutils package and prelink.
In uClibc argp functionalities has been moved from C library to
libuargp.so
Further the libc.so linker script contains an AS_NEEDED entry so that
it doesn't need to link libuargp.so explicitely.
Disable argp test if feature disabled.
Signed-off-by: Salvatore Cro <salvatore.cro@st.com>
Signed-off-by: Filippo Arcidiacono <filippo.arcidiacono@st.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Waldemar Brodkorb <wbx@uclibc-ng.org>
|
|
Commit e3c3bf2b58 introduce use of pselect6, but has following disadvantages:
* Use of userspace types in args67 structure - it breaks, for example,
configs when 32-bit uClibc-ng compiled against 64-bit kernel. Syscall
will always return EINVAL. We must use __kernel_* types and
__SYSCALL_SIGSET_T_SIZE.
* It have excess checks for NSEC_PER_SEC. Original code from select()
implementation has struct timeval => struct timespec conversion,
kernel select() syscall implementation do the same.
But none of libc versions (glibc, eglibc, musl) I know, perform similar
checks for pselect() - there is no structure fields conversions,
just struct timespec through all the calls.
To have such checks in uClibc-ng we need one example, at least.
* It is possible to avoid extra userspace reads from kernel code if
sigmask == NULL. I suggest to do it, for a few bytes cost.
* Commit didn't add test case to testsuite.
Signed-off-by: Leonid Lisovskiy <lly.dev@gmail.com>
|
|
vfork internally unwinds stack up one frame, saving the original return
address in the a3 of the unwound frame. To do this in windowed ABI it
needs to exchange two topmost bits of the original return address and of
the helper entry point. Current code doesn't do it correctly for
addresses above 0x40000000, resulting in illegal instruction exception
for configurations that put code high, e.g. for noMMU.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
|
|
Linux has a pselect syscall since 2.6.something. Using it
rather than emulating it with sigprocmask+select+sigprocmask
is smaller code, and works properly. (The emulation has
race conditions when unblocked signals arrive before or
after the select)
The tv.nsec >= 1E9 handling comes from uclibc's linux select()
implementation, which itself uses pselect() internally if the
pselect syscall exists. I though it would be good to do the
same here.
Note that although the libc pselect() API has 6 arguments,
the linux kernel syscall as 7 arguments. There is an extra,
somewhat vestigial, sizeof the signal mask argument.
Signed-off-by: Nicolas S. Dade <nic.dade@gmail.com>
|
|
Fixes compilation of some faulty applictions.
Patch from Freetz.org
Reported-by: Leonid Lisovskiy <lly.dev@gmail.com>
|
|
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
Save 20 bytes per FILE structure, avoid indirect call for
read/write/seek/close operations for normal streams.
Additionally, custom streams has fileno = -2 now, like in glibc.
bloat-o-meter report (UCLIBC_HAS_GLIBC_CUSTOM_STREAMS=y):
function old new delta
fopencookie 69 131 +62
ftello64 233 260 +27
fseeko64 298 319 +21
fclose 423 442 +19
.rodata 16696 16708 +12
fileno_unlocked 53 45 -8
__ns_name_pack 859 851 -8
vswscanf 184 144 -40
vdprintf 231 187 -44
vsscanf 210 151 -59
vswprintf 269 201 -68
vsnprintf 249 181 -68
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 5/7 up/down: 141/-295) Total: -154 bytes
Signed-off-by: Leonid Lisovskiy <lly.dev@gmail.com>
Signed-off-by: Waldemar Brodkorb <wbx@uclibc-ng.org>
|
|
In gcc 5.x there is a compile failure, for sh2 builds
as shad no longer is accepted. Strange that it is working for
4.9.x. Add back old assembly code.
|
|
Only alpha, hppa and sparc need non-default value.
|
|
glibc [BZ# 16922]
* sysdeps/mips/sys/asm.h (INT_SUB): Fix definition.
(LONG_SUB): Ditto.
(PTR_SUB): Ditto.
glibc commit 5057ad3b50
Signed-off-by: Steve Ellcey <sellcey@mips.com>
Signed-off-by: Waldemar Brodkorb <wbx@uclibc-ng.org>
|
|
Get rid of NIOS support. We try to support NIOSII.
|
|
It is marked as broken and it seems you can't get
any hardware for that anymore.
|
|
I mailed with Jan-Benedict Glaw, it seems VAX on Linux
is really a lot of work todo and uClibc support didn't work ever.
|
|
No real hardware available. The project for sh64 with sh5 seems
dead since 10 years. Gcc will remove support for it soon.
|
|
.globl can be used for every architecture so remove the define.
Sync with GNU C library.
|
|
It's even no longer required for non-ported ppc64
architecture. Sync with GNU C library.
This simplify the macros in include/libc-symbols.h.
|
|
See here for a discussion about the problem:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=47779
Latest gcc/binutils git can generate a working uClibc-ng
toolchain. Binaries tested on Arcturus uCBF54x board via
chroot from original firmware.
|
|
Actually revert 711ad9f92c1cf992c4a3d9f4f709bd692be7789c, as
it breaks vfork() on real hardware. So the common
implementation no longer works for bfin.
|
|
When installing header files a warning is generated, because
the header is mandatory.
|
|
To support older Linux kernel f.e. 2.6.32.68 include all
required definitions. Copied from musl libc fanotify.h.
Reported by Thomas Petazzoni from Buildroot project.
|
|
Implementation taken from musl libc project.
Missing functions recognized by buildroot autobuilders
with failing open-vm-tools.
|
|
|
|
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
|
|
Only on Linux alpha __NR_oldumount is defined and a umount not
umount2 syscall, but with two parameter is used.
Add special handling for it and an alias for umount2() users.
There was a discussion about this special handling, but it seems
it was never committed upstream:
http://marc.info/?l=linux-alpha&m=137455037930738&w=2
Runtime tested with qemu-alpha and a statically linked busybox
binary.
|
