On 03/22/2013 09:54 PM, Bernhard Reutner-Fischer wrote:
Bernd,
do you think that something like
http://git.openembedded.org/openembedded-core/tree/meta/recipes-devtools/gcc/gcc-4.7/gcc-uclibc-locale-ctype_touplow_t.patch
would be acceptable?
I couldn't say, I'm not a libstdc++-v3 maintainer.
On 07/10/11 07:19, Mike Frysinger wrote:
On Sat, Jul 9, 2011 at 18:38, Bernd Schmidt wrote:
So, the list of those that have an int kernel_mode_t and use
common/bits/ipc.h is x86_64 and v850. v850 has select
ARCH_LITTLE_ENDIAN and x86_64 is known to be LE as well.
looks fine then from
On 07/09/11 21:14, Mike Frysinger wrote:
On Thursday, July 07, 2011 08:04:37 Bernd Schmidt wrote:
This fixes a problem identified by Mark Salter on C6X big-endian
systems. We should not use unsigned short int mode; the proper type is
in theory mode_t, but here we need to use kernel_mode_t
This fixes a problem identified by Mark Salter on C6X big-endian
systems. We should not use unsigned short int mode; the proper type is
in theory mode_t, but here we need to use kernel_mode_t since this
structure is used for communication with the kernel, and some targets
have a 16-bit
On 07/07/11 20:41, Peter Mazinger wrote:
Hi,
This fixes a problem identified by Mark Salter on C6X big-endian
systems. We should not use unsigned short int mode; the proper
type is in theory mode_t, but here we need to use kernel_mode_t
since this structure is used for communication with the
gets remapped,
thereby invaliding anything that points to it, in particular certain
dynamic tags such as DT_NEEDED.
The following patch re-parses the dynamic info after the remapping so
as to re-compute any invalid pointers.
Signed-off-by: Bernd Schmidt ber...@codesourcery.com
---
ldso/ldso/dl-elf.c
In commit 3e0a1f388, Richard tried to fix malloc alignments by using
alignof (double __attribute_aligned__(sizeof (size_t))).
This doesn't work, since attribute_aligned overrides the alignment
rather than providing a minimum. On C6X, malloc returns four-byte
aligned values rather than the
On 04/06/2011 07:40 PM, Bernd Schmidt wrote:
In commit 3e0a1f388, Richard tried to fix malloc alignments by using
alignof (double __attribute_aligned__(sizeof (size_t))).
This doesn't work, since attribute_aligned overrides the alignment
rather than providing a minimum. On C6X, malloc returns
On 04/05/2011 11:03 PM, Bernhard Reutner-Fischer wrote:
On Fri, Apr 01, 2011 at 03:06:55PM +0200, Bernd Schmidt wrote:
+char *piclib2map = 0;
= NULL
+#ifndef MIN
+# define MIN(a,b) ((a) (b) ? (a) : (b))
MIN is always defined via sys/param.h
+cant_map:
Please use
We have a problem with DT_TEXTREL shared libraries on nommu machines.
The dynamic linker's strategy is to map the text segment read-only
first, then look for DT_TEXTREL, and use mprotect to change protections
if necessary. This fails on nommu, since a nommu kernel can decide to
share the memory
For DSBT targets (C6X only at this point), we'd like to support the case
where the user did not specify --dsbt-index at link time when building a
shared library. The dynamic linker can still assign an index at runtime
and fix up the DSBT_INDEX relocs, at the cost of startup time and memory
space.
I propose the following patch, which changes the default for
UCLIBC_UCLINUX_BROKEN_MUNMAP. That option changes malloc behaviour for
nommu systems to match munmap and mmap calls exactly, which introduces
extra overhead. Also, when I was at Analog, we've had a fair amount of
trouble making it work
On 03/05/2011 08:11 PM, Bernhard Reutner-Fischer wrote:
As to xdrproc_t, doesn't that change API WRT both manpage as well as glibc?
Yes. However, it seems to change it to a more correct form given
existing usage. There's no xdrproc I could find that's actually defined
as variadic (every example
On 03/02/2011 07:05 PM, Bernhard Reutner-Fischer wrote:
clock_nanosleep was added just recently (to master).
If you're referring to my patch, I've not committed that, as I noticed
that librt now has a version of clock_nanosleep.
Bernd
___
uClibc
I've cleaned up a few issues in the previous version of these patches,
such as those pointed out by Bernhard, a numer of whitespace issues, and
some dead code. Mark Salter has also pointed out that the
FINISH_BOOTSTRAP_RELOC macro isn't actually needed when ld.so is built
with gcc, so I've
From: Bernd Schmidt bernds_...@t-online.de ber...@codesourcery.com
This adds support for a new binary format, DSBT ELF, to the Makefiles.
Every shared library is assigned a DSBT index, and the link.so macro is
adjusted to ensure the correct linker argument is passed.
Configuration and ldso
On C6X, the stack pointer points to a word that is not part of the current
function's stack frame. It may be overwritten by callees. Take this into
account when creating the stack for a cloned thread.
Signed-off-by: Bernd Schmidt ber...@codesourcery.com
---
libpthread/linuxthreads.old
-by: Aurelien Jacquiot a-jacqu...@ti.com
Signed-off-by: Bernd Schmidt ber...@codesourcery.com
---
include/link.h | 13 -
ldso/include/dl-defs.h |2 +-
ldso/include/dl-elf.h | 11 ++-
ldso/ldso/dl-elf.c | 23
From: Mark Salter msal...@redhat.com
xdrproc_t is defined as a variadic function with two fixed arguments.
However, the code uses typecasts of non-variadic functions in assignments
to xdrproc_t types and later calls those functions as an xdrproc_t. This
leads to undefined behavior in C. On C6X,
From: Mark Salter msal...@redhat.com
This adds some DSBT index sanity checks to the runtime linker. It
catches libraries which have no index (index 0) and libraries which
try to use an already used index.
Signed-off-by: Bernd Schmidt ber...@codesourcery.com
---
ldso/ldso/dl-elf.c | 29
From: Mark Salter msal...@redhat.com
Fix spurious fall-through.
Signed-off-by: Mark Salter msal...@redhat.com
---
libc/inet/rpc/clnt_tcp.c |1 +
libc/inet/rpc/clnt_udp.c |1 +
libc/inet/rpc/clnt_unix.c |1 +
3 files changed, 3 insertions(+), 0 deletions(-)
diff --git
On 02/24/2011 12:12 AM, Mike Frysinger wrote:
On Wednesday, February 23, 2011 14:05:49 Bernd Schmidt wrote:
--- a/ldso/include/dl-defs.h
+++ b/ldso/include/dl-defs.h
@@ -212,7 +212,7 @@ typedef struct {
_dl_find_hash for this reloc TYPE. TPNT is the module in which the
matching SYM
On 02/23/2011 10:56 PM, Bernhard Reutner-Fischer wrote:
+ifeq ($(UCLIBC_FORMAT_DSBT_ELF),y)
define link.so
$(Q)$(RM) $@ $@.$(2) $(dir $@)$(1)
@$(disp_ld)
$(Q)$(CC) $(LDFLAGS-$(notdir $@)) -Wl,-soname=$(notdir $@).$(2) \
$(NOSTDLIB_CFLAGS) -o $(dir $@)$(1)
I will submit a series of patches in replies to this message which
together add a port for the Texas Instruments C6X processor family to
uClibc.
A toolchain which can be used to build it is available at
http://www.codesourcery.com/sgpp/lite/c6000/portal/release1703
We are in the process of
This adds support for a new binary format, DSBT ELF, to the Makefiles.
Every shared library is assigned a DSBT index, and the link.so macro is
adjusted to ensure the correct linker argument is passed.
Configuration and ldso support will follow in separate commits.
Signed-off-by: Bernd Schmidt
From: Mark Salter msal...@redhat.com
Fix spurious fall-through.
Signed-off-by: Mark Salter msal...@redhat.com
---
libc/inet/rpc/clnt_tcp.c |1 +
libc/inet/rpc/clnt_udp.c |1 +
libc/inet/rpc/clnt_unix.c |1 +
3 files changed, 3 insertions(+), 0 deletions(-)
diff --git
-off-by: Bernd Schmidt ber...@codesourcery.com
---
include/link.h | 13 -
ldso/include/dl-defs.h |2 +-
ldso/include/dl-elf.h | 11 ++-
ldso/ldso/dl-elf.c | 23 ++-
ldso/ldso/dl-startup.c
On C6X, the stack pointer points to a word that is not part of the current
function's stack frame. It may be overwritten by callees. Take this into
account when creating the stack for a cloned thread.
Signed-off-by: Bernd Schmidt ber...@codesourcery.com
---
libpthread/linuxthreads.old
From: Mark Salter msal...@redhat.com
xdrproc_t is defined as a variadic function with two fixed arguments.
However, the code uses typecasts of non-variadic functions in assignments
to xdrproc_t types and later calls those functions as an xdrproc_t. This
leads to undefined behavior in C. On C6X,
, otherwise provide a function that sets errno
to ENOSYS). Nathan, any input on why you chose to do it that way?
Ok to check in this version?
Signed-off-by: Bernd Schmidt ber...@codesourcery.com
Bernd
diff --git a/include/string.h b/include/string.h
index ca22055..66f64e7 100644
--- a/include
The following patch imports and adapts ffsll.c from glibc. The same
mechanism as in glibc is used to choose between ffs and ffsll to
implement ffsl.
Ok?
Bernd
Index: include/string.h
===
--- include/string.h(revision 26250)
+++
This patch, written by Nathan Sidwell, implements clock_nanosleep. Ok?
Bernd
Index: include/time.h
===
--- include/time.h (revision 307155)
+++ include/time.h (revision 307156)
@@ -349,9 +349,8 @@ extern int clock_gettime
Denys Vlasenko wrote:
+/* I think it's wrong: _toupper(c) assumes that c is a *lowercase* *letter* -
+ * it is defined as ((c) ^ 0x20)!
+ * if ((__uwchar_t) wc = 0x7f) {
+ * return (desc == _CTYPE_tolower) ? _tolower(wc) : _toupper(wc);
+ * }
+ */
This is awful. Please use #if 0 blocks,
Lennart Sorensen wrote:
On Mon, Apr 20, 2009 at 02:39:06PM -0400, Lennart Sorensen wrote:
I decided to try the blackfin version of elf2flt entirely, since it also
seems to have a fix for c++ support (not that I have any plans to use
c++), and it turns out, it is missing a single line fix in
Thierry Reding wrote:
You use git fetch to update the local repository with the latest changes
from the remote repository, then use git rebase origin/master to rebase all
local commits onto the master branch of the origin repository. origin/master
can of course be substituted with any other
Mike Frysinger wrote:
On Thursday 30 April 2009 08:57:14 Bernd Schmidt wrote:
Thierry Reding wrote:
You use git fetch to update the local repository with the latest
changes from the remote repository, then use git rebase origin/master
to rebase all local commits onto the master branch
Bernhard Reutner-Fischer wrote:
We will be switching to git (instead of svn) for our repositories.
Any particular reason?
As far as I know, git works on local repositories. How will we get
things committed into a mainline repository?
Bernd
--
This footer brought to you by insane German
Bernhard Reutner-Fischer wrote:
git push -v --thin git+ssh://$usern...@uclibc.org/srv/git/uClibc.git
Huh. Does git really work if multiple people push to the same
repository? That's news to me. All the online examples seem to suggest
using git push as a means to provide a git-pullable
Lennart Sorensen wrote:
I now wonder if there is a bug in that gcc 4.3 code, or in the linker
messing up the ordering, or possibly in the elf2flt tool.
The mainline version of elf2flt sets __[CD]TOR_LIST__ and
__[CD]TOR_END__ incorrectly. It sounds like you're hitting that
problem, so you
Bernhard Reutner-Fischer wrote:
On Wed, Feb 25, 2009 at 03:21:34PM +0100, Bernd Schmidt wrote:
Denys Vlasenko wrote:
On Wednesday 25 February 2009 01:31:01 pm Bernd Schmidt wrote:
v...@uclibc.org wrote:
reinstate __libc_foo for linuxthreads.old
The code didn't seem to have the awful #ifdefs
v...@uclibc.org wrote:
Author: vda
Date: 2009-02-25 11:15:13 + (Wed, 25 Feb 2009)
New Revision: 25450
Log:
reinstate __libc_foo for linuxthreads.old
The code didn't seem to have the awful #ifdefs before your checkins
24362 and 24363. I think it's a terrible idea to add more
Denys Vlasenko wrote:
On Wednesday 25 February 2009 01:31:01 pm Bernd Schmidt wrote:
v...@uclibc.org wrote:
reinstate __libc_foo for linuxthreads.old
The code didn't seem to have the awful #ifdefs before your checkins
24362 and 24363. I think it's a terrible idea to add more differences
Denys Vlasenko wrote:
On Monday 12 January 2009 14:07, Bernd Schmidt wrote:
Rules.mak: add -funsigned-char, to forestall future PITA
Yeah, that's just wrong for all the reasons given in the old thread.
Please revert if you haven't already.
Old thread said generated code for -fsigned-char
Denys Vlasenko wrote:
I merely plan to follow the agreed-on style in the code
I touch.
IMO for code that is just touched, following the existing style in that
file is best.
Bernd
--
This footer brought to you by insane German lawmakers.
Analog Devices GmbH Wilhelm-Wagenfeld-Str. 6
[EMAIL PROTECTED] wrote:
-sigemptyset(sa.sa_mask);
-sa.sa_flags = 0;
+/* __sigemptyset(sa.sa_mask); */
+/* sa.sa_flags = 0; */
sigaction(s, sa, NULL);
Please don't do this. Over time this will litter the code with
historical remnants and make it
Rob Landley wrote:
Can you explain what FD-PIC is in fewer words than this guy?
It's a binary format. It works on nommu, but unlike flat, it allows all
the normal ELF things like shared libraries and dynamic loading. It
needs compiler linker support so it's not something you can implement
in
Rob Landley wrote:
There are _three_ allocators right now, MALLOC, MALLOC_SIMPLE, and
MALLOC_STANDARD. Is your objection that MALLOC does everything MALLOC_SIMPLE
does (including nommu support), so there's no real need for MALLOC_SIMPLE?
Simple = plain bad in this case. It has massive
Rob Landley wrote:
So svn 23660 broke arm with my .config, but if I change my .config from
MALLOC=y to MALLOC_STANDARD=y it works again.
Does anybody understand the difference between the MALLOC
and MALLOC_SIMPLE options? The make help is not being useful here.
Off to try
Carmelo AMOROSO wrote:
No, what we do is we keep the 0.9.29 tarball around and if people have bugs
trying to use 0.9.30 they _report_ them to us. If they want to use the old
threading code, they can use the old version of the library. If they want
the new features, then they help us find
Carmelo AMOROSO wrote:
I'd like to highlight that we need to fix the testsuite build system.
I've sent a patch recently putting together some suggestions by Bernard,
Bernd and others. I'd like to have it reviewed so that it can be merged in.
I had missed that in the old thread. The patch
Denys Vlasenko wrote:
Am I reading this correctly to mean that current uclibc svn can't be used
unmodified to build uClinux-dist-2008R1-RC8 on blackfin?
Correct. I'd mentioned this before.
Bernd
--
This footer brought to you by insane German lawmakers.
Analog Devices GmbH
Daniel Jacobowitz wrote:
I have no comment on the patch itself, but I like the approach - I've
concluded before that this is the only sane way to test toolchain
pieces, especially compiler or C library. We do all of our testing
after installation here.
My patch doesn't really create a full
For gethostbyname and related functions, I can see two alternatives to
just reverting to pre-uc_malloc state, either of which would enable us
to reduce the amount of static buffers.
Looking at some other C libraries, glibc still seems to use static
buffers, but in the FreeBSD C library, there
Carmelo AMOROSO wrote:
I've now figure out that I was looking at the previous version of
string.h where all the hidden_proto
were guarded by the UCLIBC_INTERNAL.. that now is changed again.
SO my previous suggested fixed doesn't work.. so probably the only
solution is as Bernd suggested.
Denys Vlasenko wrote:
A small story. I wrote a glibc's nscd replacement last year.
glibc's nscd, when run on Google machines (and not only on them),
was crashing or locking up, and since it was broken by design,
References? I assume this has been reported and discussed somewhere?
On MMU
Bernhard Fischer wrote:
Just to make sure i understand correctly:
- If a reentrant function exists then you want the non-reentrant func to
be optional (via a central knob).
- If non-reentrant funcs are off then they are aliased to their
reentrant counterpart.
No, they have different
Denys Vlasenko wrote:
Ok, this is a scenario: the user runs a passwd utility on NOMMU box.
This utility has 20k of text and 80k of data. 70k of this data
is occupied by des.c static buffer.
At this moment the machine has only 90k RAM available, has no swap etc.
It cannot satisfy load
[EMAIL PROTECTED] wrote:
Author: vda
Date: 2008-06-06 00:25:12 -0700 (Fri, 06 Jun 2008)
New Revision: 22244
Log:
Use more natural way to return a value in __read_etc_hosts_r:
- if (buflen sizeof(char *)*(ALIAS_DIM))
+ if (buflen sizeof(char *)*ALIAS_DIM)
return
[EMAIL PROTECTED] wrote:
Remove superfluous attribute_hidden from function
definition. I seems to produce spurious warning:
libc/inet/resolv.c:1549: warning: 'visibility' attribute ignored on non-class
types
(seems like gcc bug)
-FILE * attribute_hidden __open_etc_hosts(void)
+FILE *
Bernhard Fischer wrote:
PS:
You know that i usually do not blindly rewrite stuff, but the resolver
code is really an ugly and bloated mess (and doesn't even work for
IPv6-only), but this time i think it would be saner to write this thing
from scratch and not try to beat it into shape.
Surely
Bernhard Fischer wrote:
Author: vda
Date: 2008-06-06 00:25:12 -0700 (Fri, 06 Jun 2008)
New Revision: 22244
Log:
Use more natural way to return a value in __read_etc_hosts_r:
hah. vda bashing! ;) No, serious:
The proper term is patch review. I've gone through the last 200
checkins
[EMAIL PROTECTED] wrote:
+ libpthread/*/*.$(ext) libpthread/*/*/*.$(ext) \
+ libpthread/*/*/*/*.$(ext) libpthread/*/*/*/*/*.$(ext) \
+ libpthread/*/*/*/*/*/*.$(ext) libpthread/*/*/*/*/*/*/*.$(ext) \
+ libpthread/*/*/*/*/*/*/*/*.$(ext))
Uh... is
[EMAIL PROTECTED] wrote:
Author: vda
Date: 2008-05-20 13:32:27 -0700 (Tue, 20 May 2008)
New Revision: 22027
Log:
getopt: do not needlessly use static structure.
Reorder structure members and change some of them into smallints
to reduce bss and text:
textdatabss dec hex
Denys Vlasenko wrote:
I want to do it in Linux from scratch style. This way,
I will get better understanding of the way it is done,
and possibly stumble upon rough edges in busybox,
uclubc, etc. This will lead to itch scratching
and bug fixing.
It will also lead to unnecessary pain.
elf
There appears to be one additional problem. Sometimes I get the
following cycle:
[EMAIL PROTECTED] /local/src/uclibc/bernds/uClibc $ make
make[1]: `conf' is up to date.
STRIP -X --strip-debug -R .note -R .comment lib/libpthread.a
AR cr lib/libpthread.a
[EMAIL PROTECTED]
Denys Vlasenko wrote:
Are you building blackfin toolchain(s) too?
Can you let me know exact version of binutils you use,
and also full configure command line which you use?
Do you set some special CFLAGS etc?
Just use the buildscript, please. That ensures you get the same
toolchain that
Bernhard Fischer wrote:
- adds several config-options to allow for turning off certain features
This seems a good way to reduce size.
One additional idea I had recently was to add a config option to turn
off non-reentrant functions where a reentrant version exists (e.g.
gethostbyname vs
Denys Vlasenko wrote:
Thanks. The attached patch fixes it (I applied it to uclibc svn).
These patches are not self-explanatory. Could you provide some
information about the cause of the failures, and how your patches fix them?
Bernd
--
This footer brought to you by insane German lawmakers.
Denys Vlasenko wrote:
None of include files are tracked, though.
find -name '*.h' | xargs touch + make only runs oldconfig
but does not rebuild any object files.
That seems like a better reason to change things around.
Bernd
--
This footer brought to you by insane German lawmakers.
Analog
Denys Vlasenko wrote:
The new system I have worked on simplifies these Makefiles.
A lot of this is just the klib-y trick, which could be done just as well
in the existing system.
I don't think simplifying the Makefiles should be a goal by itself, as
they aren't all that complicated - maybe a
Denys Vlasenko wrote:
As I said, I am not obsessed with build system replacement.
I started it as a necessity because I was bitten several times with
the modification of files not being detected.
I thought that it is a sign that build system is not maintained
and is bit rotting.
After recent
[EMAIL PROTECTED] wrote:
- if (!(tpnt-init_flag RELOCS_DONE))
+ if (!(tpnt-init_flag RELOCS_DONE)) {
_dl_if_debug_dprint(relocation processing: %s\n,
tpnt-libname);
-
+ }
Seems to me it would be better to fix the macro than all of its callers?
Bernd
--
This
Bernhard Fischer wrote:
So the only immediately annoying thing in ldso remains:
ldso/ldso/dl-elf.c: In function '_dl_load_shared_library':
ldso/ldso/dl-elf.c:201: warning: unused parameter 'tpnt'
The solution in the gcc tree is to use __attribute__((unused)) for
anything that may be unused
Denys Vlasenko wrote:
I noticed that uclibc build system is broken.
After I touch a file, running make won't rebuild it.
As far as I can tell, only files that go into libc.so and not into
libc.a, such as forward.c from libpthread, are affected. This seems to
have been broken by revision
Denys Vlasenko wrote:
Before I apply this, I wanted to start a discussion about whether
__uc_malloc is a good idea at all. Space savings are all well and good,
but these come at a cost in reliability.
There is no cost in reliability.
Face it: if you have no free memory - you have no free
A number of programs, such as busybox, contain calls to fork which may
not necessarily get executed at runtime. On nommu systems, this
currently produces link errors since the function doesn't exist.
In our Blackfin tree, we use the following patch (by Mike IIRC) to
convert the link error into a
Carmelo AMOROSO wrote:
based on the patch from Nickolai, here you can find a comprehensive
patch to fix
dladdr function.
With the current implementation, the invocation of dladdr((void *) 1,
dlinfo)
will fill dlinfo.dli_fname with the name of the application itself and
dlinfo.dli_fbase
For the Blackfin, we need to compile several versions of uClibc and
install them in such a way that the cross compilers will find them. To
do this, we've come up with the following patch (combined from multiple
contributions by Jie Zhang, Mike Frysinger and myself).
Essentially, this adds two
Daniel Jacobowitz wrote:
We don't have any lib/$(MULTILIB) directories; it was much simpler to
put each into its own sysroot and it works nicely at runtime too.
Okay.. what kind of multilibs do you have, and what's your directory
structure?
Maybe Mike or Jie could comment on why exactly we use
Carmelo AMOROSO wrote:
based on the patch from Nickolai, here you can find a comprehensive
patch to fix
dladdr function.
With the current implementation, the invocation of dladdr((void *) 1,
dlinfo)
will fill dlinfo.dli_fname with the name of the application itself and
dlinfo.dli_fbase
Carmelo AMOROSO wrote:
I think that when nptl merge will be completed, we could use something
like that:
#if defined USE_TLS || defined __FDPIC__
#define HASH_EXTRA_TPNT
#else
#undef HASH_EXTRA_TPNT
#endif
and use it in _dl_find_hash wrapper.
I'd much rather get rid of the ifdeffery
No one made further comments regarding the second set of patches, so
I've now made another pass over them to fix a few problems and style
issues, and checked them in.
A few minor issues remain to be resolved to get the Blackfin support
100% working, and the FRV needs attention as well. I'll be
Carmelo Amoroso wrote:
with the solution proposed into the nptl branch, you can keep the caller
of _dl_find_hash always the same. Where the extra tpnt parameter is not
required,
independently from TLS or FDPIC code, like into ldso.c to lookup
some function (malloc), you can safely pass NULL
Carmelo Amoroso wrote:
please have a look at the nptl branch where already is the updated
_dl_find_hash
for NPTL using the same extra parameter you need. There is also a wrapper
around it as suggested by Jocke.
I think there is a way to use the interface for nptl into your case.
In these
Joakim Tjernlund wrote:
The 02_underscores patch is really ugly.
Any suggestions how to improve it?
remove _ syms from gcc would be best. Do you really need them?
Yes. The assembly language isn't really parseable otherwise.
Don't like the dl_find_hash_mod v.s dl_find_hash thingy.
What do
Joakim Tjernlund wrote:
-Original Message-
From: Bernd Schmidt [mailto:[EMAIL PROTECTED]
Joakim Tjernlund wrote:
The 02_underscores patch is really ugly.
Any suggestions how to improve it?
remove _ syms from gcc would be best. Do you really need them?
Yes. The assembly language
These are the FD-PIC patches necessary to get ld.so working on the
Blackfin. These were originally written by Alex Oliva for the FRV and
updated to work with current uClibc sources by myself.
I'm posting these for review; I intend to check them in in a few days if
there are no objections.
Note
Joakim Tjernlund wrote:
Did a quick scan of the patches and came up with these comments:
While they tell me what you dislike, they don't really provide much in
the way of alternatives.
The 02_underscores patch is really ugly.
Any suggestions how to improve it?
Don't like the
Joakim Tjernlund wrote:
-Original Message-
From: Bernd Schmidt [mailto:[EMAIL PROTECTED]
Sent: den 24 juni 2007 16:42
To: Joakim Tjernlund
Cc: 'uClibc'
Subject: Re: libpthread vs libc dynamic link order
Joakim Tjernlund wrote:
I never got the hang of the hidden_{proto,def} stuff
We have the following patches in our tree to improve shared flat support
in uClibc for the Blackfin.
* We want to use -mleaf-id-shared-library to compile libc (but only
libc). This works best by setting PICFLAG to -mleaf-id-shared-library
and using libc_so.a instead of libc.a in the link.
Koen Kooi wrote:
Bernd Schmidt schreef:
We have the following patches in our tree to improve shared flat support
in uClibc for the Blackfin.
* We want to use -mleaf-id-shared-library to compile libc (
Which upstream gcc version(s) support that flag?
This is only for the Blackfin, where
Joakim Tjernlund wrote:
I never got the hang of the hidden_{proto,def} stuff. I think these
are way overcomplex.
I don't see anything complex about hidden_proto/def. All they do is
create a hidden version of the symbol (e.g. __GI_pthread_cond_signal)
and an exported alias
92 matches
Mail list logo
