Age | Commit message (Collapse) | Author |
|
This optimization is based on prefetching and 64bit data transfer via FPU
(only for the little endianess)
Tests shows that:
----------------------------------------
Memory bandwidth | Gain
| sh4-300 | sh4-200
----------------------------------------
512 bytes to 16KiB | ~20% | ~25%
from 32KiB to 16MiB | ~190% | ~5%
----------------------------------------
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
|
|
Conflicts:
Rules.mak
libc/misc/sysvipc/msgq.c
test/Rules.mak
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
This avoids a warnings about implicit declaration of function 'memset'.
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
[__]isascii need to be defined all the time for the build.
Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Move attribute_noreturn to header.
Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
We don't provide spawn.h let alone any other spawn funcs/types, so don't
set up the _POSIX_SPAWN define that some packages (like vlc) check.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
Call the hidden memcpy() ourselves otherwise gcc will emit a call to the
public memcpy() which goes through the PLT.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
Assign each field one by one rather than stack initialization as gcc will
call memset() to zero out the rest of the structure -- which we don't care
about as the field is unused and not seen outside of the libc.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
We only need to set the first byte to 0, but gcc likes to zero out the
rest of the string with memset() when using this initialization style.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
This also has the advantage of fputs() having a hidden alias while
puts does not.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
__gen_tempname now needs to not be hidden so libpthread can get at it
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
sem_open(3) needs to create a temporary file in a way which can't
be efficiently implemented in terms of POSIX API. Extend
__gen_tempname with mode_t mode argument in order to ease
sem_open implementation.
Signed-off-by: Mikhail Gusarov <dottedmag@dottedmag.net>
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
We don't provide spawn.h let alone any other spawn funcs/types, so don't
set up the _POSIX_SPAWN define that some packages (like vlc) check.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
Call the hidden memcpy() ourselves otherwise gcc will emit a call to the
public memcpy() which goes through the PLT.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
Assign each field one by one rather than stack initialization as gcc will
call memset() to zero out the rest of the structure -- which we don't care
about as the field is unused and not seen outside of the libc.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
We only need to set the first byte to 0, but gcc likes to zero out the
rest of the string with memset() when using this initialization style.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
This also has the advantage of fputs() having a hidden alias while
puts does not.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
* setup memory for the cancellation buffer
* remove unneeded weak's
* deallocation of tsd on main thread exit
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
* clock_getcpuclockid, clock_gettime, clock_nanosleep, mq_receive,
mq_send, mq_timedreceive, mq_timedsend, _SC_MONOTONIC_CLOCK
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
* SAVE_PID, RESTORE_PID in vfork.S
* clone.S tweaks to allow for the pid to be reset
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
* RESET_PID support for clone.S
* SAVE_PID, RESTORE_PID in vfork.S
* fixup syscall assembly constraints to be a little less restrictive
allows arbitrary variables to be used as the syscall nr
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
* vfork impl with support for SAVE_PID, RESTORE_PID
* sysdep.h assembly macros
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
* unified atomic.h compare and exchange macros
* clone.S with RESET_PID support
* sh specific versions of pread/pwrite with cancellation support
* check SHARED instead of PIC
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
* a tweaked clone.S with RESET_PID support
* atomic.h with cmpxchg that works for v8 and beyond
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
* updated kernel-features.h
* system is provided by pt-system with nptl
* _exit should do exit_group with nptl
* tsd tls ptr in libc
* rt_sigwaitinfo impl added
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
|
|
The malloc() code checks the incoming size to make sure the header
adjustment doesn't cause overflow in the size storage. Add the same
check to realloc() to catch stupid stuff like realloc(..., -1).
Reported-by: James Coleman <james.coleman@ubicom.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
We don't really need to know the exact symbol that caused a failure as
long as we know where to start looking. So unify the duplicate code
between all funcs and between the sparc variants. This gives us a nice
code shrink of ~95%.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
The common define has the same behavior, just formatted differently.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
Recent SH kernel headers merge 32-bit and 64-bit headers, changing the
include guards on asm/posix_types.h in the process; update uClibc code
depending on those include guards.
Signed-off-by: Joseph Myers <joseph@codesourcery.com>
|
|
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
|
|
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
|
|
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
LEGACY was removed for utimes() in SUSv4
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Apps should switch to nftw()
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|