| Age | Commit message (Collapse) | Author |
|---|
|
include features.h to make the next ifdef work.
Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>
|
|
Use ADDR_ALIGN to align the minvma when loading shared libraries instead
of the hard coded 0xffffU value.
This fixes teh stand/alone support on ARM as reported in bug #3133.
Signed-off-by: Sven Ola <sven-ola@gmx.de>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
|
|
ARM disables by default the support for ASM CFI directives.
Anyway using an old version of binutils that does not support some new
pseudo-op, the build fails as below:
AS libpthread/nptl/sysdeps/unix/sysv/linux/close.oS
libpthread/nptl/sysdeps/unix/sysv/linux/close.S: Assembler messages:
libpthread/nptl/sysdeps/unix/sysv/linux/close.S:9: Error: unknown pseudo-op: `.cfi_sections'
The problem is that the .cfi_sections pseudo should be wrapped by a macro
that expands to nothing when the CFI is off.
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
CC: Khem Ray <ray.khem@gmail.com>
|
|
Because the nommu address space is flat, and the application stack can
literally be located anywhere, we cannot rely on the assumptions that the
mmu port gets away with. Namely, that the first thread's stack lives at
the top of memory and nothing will be created above it.
Currently, the code rounds the current stack up a page and sets that as
the "top" of the stack, and then marks the "bottom" of the stack as "1".
Then as new threads are created, this assumption is further refined by
slowly backing off the "bottom" when a new stack is created within the
range of the initial stack.
Simple ascii example (tid0 is the initial thread):
1 thread:
[bos tid0 stack tos]
2 threads:
[ tid0 stack ]
[tid1 stack]
3 threads:
[ tid0 stack ]
[tid1 stack]
[tid2 stack]
As you can kind of see, this algorithm operates on one basic assumption:
the initial top of stack calculation is the absolute top of the stack.
While this assumption was fairly safe in the original nommu days of yore
where the only file format was FLAT (which defaults to a 4KiB stack --
exactly 1 page), and memory was fairly tight, we can see that this falls
apart pretty quickly as soon as the initial stack is larger than a page.
The issue that crops up now is simple to hit: start an application with
an 8KiB stack, execute some functions that put pressure on the stack so
that it exceeds 4KiB, then start up some threads. The initial tos will
be rounded up by a page, but this is actually the middle of the stack.
Now when the initial thread returns from its functions (thus unwinding
the stack) and tries to call something which calls back into libpthread,
the thread_self() func fails to detect itself as the initial thread as
the current stack is now above the tos. The __pthread_find_self() func
kicks in, walks all the thread arrays, fails to find a hit, and then
walks into uninitialized memory for the thread descriptor. Use of this
garbage memory has obvious results -- things fall down & go boom.
To address this, I extend the current algorithm to automatically scale
back both the bottom and the top stack limits of the initial thread.
We use the current stack pointer at "thread boot time" only as a single
known point. The initial thread stack bottom is set to the bottom of
memory and the initial thread stack top is set to the top of memory.
Then as we create new stack threads, we figure out whether the new stack
is above or below the single known good address, and then scale back
either the tos or the bos accordingly.
Reviewed-by: Steven J. Magnani <steve@digidescorp.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
We want to access the frame pointer, so do so directly rather than
"overflowing" the pregs array and ending up at the fp member.
This fixes the Blackfin build warnings:
libpthread/linuxthreads.old/ptlongjmp.c: In function 'pthread_cleanup_upto':
libpthread/linuxthreads.old/ptlongjmp.c:35: warning: array subscript is above array bounds
libpthread/linuxthreads.old/ptlongjmp.c:56: warning: array subscript is above array bounds
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
The printf precision takes an integer, not a size_t. Otherwise we get:
libc/misc/internals/tempname.c: In function '___path_search':
libc/misc/internals/tempname.c:116: warning:
field precision should have type 'int', but argument 3 has type 'size_t'
field precision should have type 'int', but argument 5 has type 'size_t'
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Commit 33cb7f0b4 tried to add a small optimization for skipping unnecessary
.dynamic adjustments, but did so by referencing an opaque type. While this
works for non-fdpic targets (since the type can be cast to an integer), it
falls apart for fdpic targets where the type is actually a structure.
Since FDPIC can't support this optimization without walking a series of
linked structures, just skip it.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
Newer gcc's will generate a call to cacheflush when updating jump tables,
and that has to be done in kernel space (to avoid hardware anomalies). So
make sure uClibc provides that symbol.
Signed-off-by: Steve Kilbane <steve@whitecrow.demon.co.uk>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
Once we pull in the header, we're forced to declare the syscall with
all the right types.
Signed-off-by: Steve Kilbane <steve@whitecrow.demon.co.uk>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Somehow Makefile.in ended in there, remove it.
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: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
0f85b228 used 'filter-pout' instead of 'filter-out'. Fix that.
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
The linked list of library dependencies created by dlopen() was not
being freed by dlclose().
Signed-off-by: Philip Craig <philipjcraig@gmail.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Commits 1f6601a and 094d82d introduced the "else ifeq" construct, which
requires GNU make v3.81 or higher. This breaks the build on RHEL4 hosts.
Signed-off-by: Kevin Cernekee <cernekee@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: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
See GCC PR47577; TODO: Remove them.
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
When a string format is processed and the argument is NULL, this yields
"(null)" regardless of precision. This does not make sense, precision
should not be exceeded. A simple test shows that glibc outputs nothing
if precision is smaller than six and the attached patch implements this
same behaviour.
Consider the not uncommon case of strings implemented like this:
struct string { int len; char *ptr; };
There is often no nultermination and they may be printed like this:
printf("%.*s", string.len, string.ptr);
If len is 0 then ptr may be anything, but NULL is a common value.
Obviously the empty string would be expected, not "(null)".
Signed-off-by: Jones Desougi <jones.desougi@27m.se>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
A number of new address / protocol families have been added over the
years, so sync with Linux 2.6.38-rc3, adding CAN, ISDN, Phonet, Zigbee, ..
which are starting to be used by applications.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Thanks to Nitin Garg for notincing!
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
|
|
Signed-off-by: Nitin Garg <nitingarg98@gmail.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
|
|
Nios II is not i386 :-)
Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
|
|
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
Signed-off-by: Richard Braun <rbraun@sceen.net>
Acked-by: Joakim Tjernlund <joakim.tjernlund@transmode.se>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
Fix libc.so linker script generation to add libubacktrace AS_NEEDED entry
using MULTILIB_DIR instead of hard-coded "lib".
Signed-off-by: Filippo Arcidiacono <filippo.arcidiacono@st.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
|
|
harmless copy'n paste error in #error
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
|
|
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
|
|
The current C macro magic does not correctly concatenate the
__USER_LABEL_PREFIX__ string on architectures where it has a
non-empty value. We need to use __stringify to get the desired
behaviour.
Signed-off-by: Will Newton <will.newton@imgtec.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
|
|
This patch adds a working implementation of pthread_create for architectures
where STACK_GROWS_UP.
Signed-off-by: Matt Fleming <matthew.fleming@imgtec.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
|
|
Now _dl_find_hash and _dl_lookup_hash are exactly the same, we can get rid
of the _dl_lookup_hash, reverting the _dl_find_hash from a wrapper of
_dl_lookup_hash to its original role.
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
|
|
Protected symbols are global symbols for which interposition is not allowed.
We manage them in generic _dl_lookup_hash function. To handle protected symbols
we need to get a reference to the module that defines the symbol itself.
So we pass a new parameter 'struct symbol_ref' to the __dl_lookup_hash
that is defined as below:
struct symbol_ref {
const ElfW(Sym) *sym;
struct elf_resolve *tpnt;
};
The tpnt field is used as an ouput parameter and refers to the module which defines
the protected symbol.
Further it can be used as output parameter for TLS relocations and FDPIC case.
The sym field is instead used as an input parameter to detect the visibility of the
symbol we are looking-up.
In this way we get rid of different signatures for _dl_lookup_hash, allowing to remove
the _dl_find_hash wrapper.
This new structure is also suitable for prelink integration.
Signed-off-by: Salvatore Cro <salvatore.cro@st.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
|
|
This reverts commit 48fb264beaac8114e5ac3e80e70dda473fbce96d.
The generic implementation will cover all the architectures handling
the protected symbols in _dl_lookup_hash [ldso/ldso/dl-hash.c]
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
|
|
This reverts commit 74407db52d3953c7f3c6b8a53661cfc96cb07e22.
The generic implementation will cover all the architectures handling
the protected symbols in _dl_lookup_hash [ldso/ldso/dl-hash.c]
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
|
|
This reverts commit ba38f0cec27b91cc7c605417ad047c4dc77d732f.
The generic implementation will cover all the architectures handling
the protected symbols in _dl_lookup_hash [ldso/ldso/dl-hash.c]
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
|
|
Thanks Waldemar Brodkorb for reporting
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
It's been found that on uclibc-based system most of the programs log with
'user' facility despite the calls to openlog() with other facilities explicitly
set. Most but not all. Investigating that I've found that the ones logging with
proper facilities used LOG_NDELAY in openlog(). The problem is that LogFacility
is overwritten in openlog_intern and LOG_USER is passed to it from vsyslog() no
matter what the program set with openlog().
Signed-off-by: Roman I Khimov <khimov@altell.ru>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
closes bug #3037
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
The generation of the sysnum.h has to be a prerequisite of
'pregen' target instead of headers, because it needs the compiler
already available.
The so called C bootstrap headers are required to build the compiler
by invoking 'make headers', so the 'headers' target cannot relies itself
on the compiler to be invoked.
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
|
|
According to git man page the syntax for git archive should have
the <tree-ish> as last argument. Indeed using git 1.5.5.6 the tarball
was not properly created. Newer version (i.e. 1.6.3.3) works fine even if
HEAD comes before other option. For backward compatibility use a syntax
that is fine with older git.
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
|
|
defined
A missing change in the previous commit.
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
|
