Nabble - Sourceware - binutils list

Re: Help about prelinker on mips

2009-12-22T06:36:54Z

On Mon, Dec 21, 2009 at 10:22 AM, Amker.Cheng <amker.cheng@...> wrote:

> Thanks very much.
> After applied the two patches (both from ) in :
> http://sourceware.org/ml/prelink/2008-q4/msg00001.html
> the prelinker complained new messages like :
> -----------------------
> Assigned virtual address space slots for libraries:
> /lib/ld.so.1 2c000000-2c02fa44
> /lib/libc.so.6 2c040000-2c1a5b80
> /lib/libdl.so.2 2c1b0000-2c1c20e0
> /lib/libcrypt.so.1 2c1d0000-2c2101fc
> /lib/libnsl.so.1 2c220000-2c2478b0
> /lib/libresolv.so.2 2c250000-2c263960
> /lib/libutil.so.1 2c270000-2c2820d0
> Prelinking /mnt/nfs/tools/dummy
> ./prelink: /mnt/nfs/tools/dummy: Could not find symbol copy reloc is against
>
> -----------------------
> What does this mean? I found only one post about this message:
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=411356
>
> and it seems something wrong with the application, rather than tools.

I rechecked the patches,the "Could not find symbol copy reloc" error
was gone if
not patch "http://sourceware.org/ml/prelink/2008-q4/msg00006.html".
Now it works with patch from
http://sourceware.org/ml/prelink/2008-q4/msg00001.html,

Thanks you again.

--
Best Regards.

Re: gold -v should continue and do the link just like GNU ld.

2009-12-21T17:28:06Z

Sriraman Tallam <tmsriram@...> writes:

> * options.cc (General_options::parse_version): Make -v continue and do
> the link like GNU ld does.

This is OK.

Thanks.

Ian

Re: [Patch ARM] Fix b / bl ranges for Thumb2.

2009-12-21T17:03:46Z

On Mon, 2009-12-21 at 12:02 +0000, Ramana Radhakrishnan wrote:

> Hi,
>
> In stage1 of a bootstrap of GCC in Thumb2 state, I came across a problem
> with unconditional calls going out of range in Thumb2. This was because
> the assembler ends up with encoding a bl within the function with the
> Thumb1 bl instruction rather than the Thumb2 instruction. With some help
> from Richard, I've now fixed this with the attached patch.
>
> This refactors some of the code from md_apply_fix into a separate
> function of its own and rewrites some of the masking code in a form more
> friendly to immediate generation on the ARM.
>
> Tested cross on arm-eabi with no regressions. Ok to commit on trunk and
> backport to the 2.20 branch ?
>
> cheers
> Ramana
>
> 2009-12-21 Ramana Radhakrishnan <ramana.radhakrishnan@...>
> Richard Earnshaw <richard.earnshaw@...>
>
> * config/tc-arm.c (encode_thumb2_b_bl_offset): New. Refactored
> from md_apply_fix.
> (md_apply_fix): Fixup range checks for Thumb2 version
> of unconditional calls. Call encode_thumb2_b_bl_offset for
> unconditional branches / function calls.
>
>
>

Please also add a regression test for this case.

R.

gold -v should continue and do the link just like GNU ld.

2009-12-21T15:39:49Z

Hi,

GNU ld does go ahead and link after -v and does not stop after
printing the version. Here is a patch to make gold do the same.

* options.cc (General_options::parse_version): Make -v continue and do
the link like GNU ld does.

Index: options.cc
===================================================================
RCS file: /cvs/src/src/gold/options.cc,v
retrieving revision 1.94
diff -u -u -p -r1.94 options.cc
--- options.cc 14 Oct 2009 05:25:01 -0000 1.94
+++ options.cc 21 Dec 2009 23:27:26 -0000
@@ -288,8 +288,10 @@ General_options::parse_help(const char*,
void
General_options::parse_version(const char* opt, const char*, Command_line*)
{
- gold::print_version(opt[0] == '-' && opt[1] == 'v');
- ::exit(EXIT_SUCCESS);
+ bool print_short = (opt[0] == '-' && opt[1] == 'v');
+ gold::print_version(print_short);
+ if (!print_short)
+ ::exit(EXIT_SUCCESS);
}

Note that, --version should just stop after printing the message.

Is this ok to commit ?

Thanks,
-Sriraman.

Re: [Patch ARM] Fix b / bl ranges for Thumb2.

2009-12-21T04:58:53Z

Thanks Nick,

> This should be one line, not two.

I've committed with that change adjusted for line lengths. Can I confirm
that this is ok to backport to the 2.20 branch as well ?

cheers
Ramana
>
> Cheers
> Nick

Re: [Patch ARM] Fix b / bl ranges for Thumb2.

2009-12-21T04:39:39Z

Hi Ramana,

> 2009-12-21 Ramana Radhakrishnan <ramana.radhakrishnan@...>
> Richard Earnshaw <richard.earnshaw@...>
>
> * config/tc-arm.c (encode_thumb2_b_bl_offset): New. Refactored
> from md_apply_fix.
> (md_apply_fix): Fixup range checks for Thumb2 version
> of unconditional calls. Call encode_thumb2_b_bl_offset for
> unconditional branches / function calls.

Approved - please apply.

There is one small formatting nit:

+ else
+ if ((value & ~0x1ffffff) && ((value & ~0x1ffffff) != ~0x1ffffff))

This should be one line, not two.

Cheers
Nick

[Patch ARM] Fix b / bl ranges for Thumb2.

2009-12-21T04:02:01Z

Hi,

In stage1 of a bootstrap of GCC in Thumb2 state, I came across a problem
with unconditional calls going out of range in Thumb2. This was because
the assembler ends up with encoding a bl within the function with the
Thumb1 bl instruction rather than the Thumb2 instruction. With some help
from Richard, I've now fixed this with the attached patch.

This refactors some of the code from md_apply_fix into a separate
function of its own and rewrites some of the masking code in a form more
friendly to immediate generation on the ARM.

Tested cross on arm-eabi with no regressions. Ok to commit on trunk and
backport to the 2.20 branch ?

cheers
Ramana

2009-12-21 Ramana Radhakrishnan <ramana.radhakrishnan@...>
Richard Earnshaw <richard.earnshaw@...>

* config/tc-arm.c (encode_thumb2_b_bl_offset): New. Refactored
from md_apply_fix.
(md_apply_fix): Fixup range checks for Thumb2 version
of unconditional calls. Call encode_thumb2_b_bl_offset for
unconditional branches / function calls.

commit a87e23ee88b01c224e8df3628e290f134d4fe3fd
Author: Ramana Radhakrishnan <ramrad01@...>
Date: Mon Dec 21 11:05:57 2009 +0000

Commit Fix for bl and b encoding for Thumb2

diff --git a/gas/config/tc-arm.c b/gas/config/tc-arm.c
index 65d81a9..01e9f3a 100644
--- a/gas/config/tc-arm.c
+++ b/gas/config/tc-arm.c
@@ -19532,6 +19532,31 @@ arm_optimize_expr (expressionS *l, operatorT op, expressionS *r)
return FALSE;
}

+/* Encode Thumb2 unconditional branches and calls. The encoding
+ for the 2 are identical for the immediate values. */
+
+static void
+encode_thumb2_b_bl_offset (char * buf, offsetT value)
+{
+#define T2I1I2MASK ((1 << 13) | (1 << 11))
+ offsetT newval;
+ offsetT newval2;
+ addressT S, I1, I2, lo, hi;
+
+ S = (value >> 24) & 0x01;
+ I1 = (value >> 23) & 0x01;
+ I2 = (value >> 22) & 0x01;
+ hi = (value >> 12) & 0x3ff;
+ lo = (value >> 1) & 0x7ff;
+ newval = md_chars_to_number (buf, THUMB_SIZE);
+ newval2 = md_chars_to_number (buf + THUMB_SIZE, THUMB_SIZE);
+ newval |= (S << 10) | hi;
+ newval2 &= ~T2I1I2MASK;
+ newval2 |= (((I1 ^ S) << 13) | ((I2 ^ S) << 11) | lo) ^ T2I1I2MASK;
+ md_number_to_chars (buf, newval, THUMB_SIZE);
+ md_number_to_chars (buf + THUMB_SIZE, newval2, THUMB_SIZE);
+}
+
void
md_apply_fix (fixS * fixP,
valueT * valP,
@@ -20224,10 +20249,6 @@ md_apply_fix (fixS * fixP,
fixP->fx_r_type = BFD_RELOC_THUMB_PCREL_BRANCH23;
#endif

- if ((value & ~0x3fffff) && ((value & ~0x3fffff) != ~0x3fffff))
- as_bad_where (fixP->fx_file, fixP->fx_line,
- _("branch out of range"));
-
if (fixP->fx_r_type == BFD_RELOC_THUMB_PCREL_BLX)
/* For a BLX instruction, make sure that the relocation is rounded up
to a word boundary. This follows the semantics of the instruction
@@ -20235,17 +20256,25 @@ md_apply_fix (fixS * fixP,
1 of the base address. */
value = (value + 1) & ~ 1;

- if (fixP->fx_done || !seg->use_rela_p)
- {
- offsetT newval2;

- newval = md_chars_to_number (buf, THUMB_SIZE);
- newval2 = md_chars_to_number (buf + THUMB_SIZE, THUMB_SIZE);
- newval |= (value & 0x7fffff) >> 12;
- newval2 |= (value & 0xfff) >> 1;
- md_number_to_chars (buf, newval, THUMB_SIZE);
- md_number_to_chars (buf + THUMB_SIZE, newval2, THUMB_SIZE);
+ if ((value & ~0x3fffff) && ((value & ~0x3fffff) != ~0x3fffff))
+ {
+ if (!(ARM_CPU_HAS_FEATURE (cpu_variant, arm_arch_t2)))
+ {
+ as_bad_where (fixP->fx_file, fixP->fx_line,
+ _("branch out of range"));
+ }
+ else
+ if ((value & ~0x1ffffff) && ((value & ~0x1ffffff) != ~0x1ffffff))
+ {
+ as_bad_where (fixP->fx_file, fixP->fx_line,
+ _("Thumb2 branch out of range"));
+ }
}
+
+ if (fixP->fx_done || !seg->use_rela_p)
+ encode_thumb2_b_bl_offset (buf, value);
+
break;

case BFD_RELOC_THUMB_PCREL_BRANCH25:
@@ -20254,26 +20283,8 @@ md_apply_fix (fixS * fixP,
_("branch out of range"));

if (fixP->fx_done || !seg->use_rela_p)
- {
- offsetT newval2;
- addressT S, I1, I2, lo, hi;
+ encode_thumb2_b_bl_offset (buf, value);

- S = (value & 0x01000000) >> 24;
- I1 = (value & 0x00800000) >> 23;
- I2 = (value & 0x00400000) >> 22;
- hi = (value & 0x003ff000) >> 12;
- lo = (value & 0x00000ffe) >> 1;
-
- I1 = !(I1 ^ S);
- I2 = !(I2 ^ S);
-
- newval = md_chars_to_number (buf, THUMB_SIZE);
- newval2 = md_chars_to_number (buf + THUMB_SIZE, THUMB_SIZE);
- newval |= (S << 10) | hi;
- newval2 |= (I1 << 13) | (I2 << 11) | lo;
- md_number_to_chars (buf, newval, THUMB_SIZE);
- md_number_to_chars (buf + THUMB_SIZE, newval2, THUMB_SIZE);
- }
break;

case BFD_RELOC_8:

Re: Binutils Internals Document In PDF Format.

2009-12-21T03:53:04Z

Nick Clifton wrote:

> Hi Prashant,
>
>> I already read that document but, I need in depth information about
>> porting Binutils like in case of GCC, we have 'GCC internals document'
>
> There are a couple of internals documents in the binutils sources. See:
>
> bfd/doc/bfdint.texi
> ld/ldint.texinfo
> gas/doc/internals.texi

Or run "make pdf" in your latest build dir (needs pdftex et al.)

cheers,
DaveK

Re: Binutils Internals Document In PDF Format.

2009-12-21T03:22:15Z

Hi Prashant,

> I already read that document but, I need in depth information about
> porting Binutils like in case of GCC, we have 'GCC internals document'

There are a couple of internals documents in the binutils sources. See:

bfd/doc/bfdint.texi
ld/ldint.texinfo
gas/doc/internals.texi

But in the main the documentation is the sources themselves. Most ports
start by copying the code of an already existing, similar port, and then
modifying it for the needs of the new target.

Cheers
Nick

Re: Binutils Internals Document In PDF Format.

2009-12-21T03:04:45Z

Hi Nick,

Thanks for your reply.

I already read that document but, I need in depth information about
porting Binutils like in case of GCC, we have 'GCC internals document'
which specifies all the things about porting GCC to new architecture.

If you know any such document, please send me as I am unable to find
any and also, I referred the reference documents mentioned on the same
link you sent.

Thanks & Regards,
Prashant.

On Mon, 2009-12-21 at 10:53 +0000, Nick Clifton wrote:

> Hi Prashant,
>
> > I am new to Binutils and need the internals document ( Better if in
> > 'PDF' format. ) so as to port Binutils to a new architecture.
>
> A binutils contributor has recently made just such a document, although
> it is in a plain text format rather than PDF. You can find it here:
>
> http://sourceware.org/binutils/binutils-porting-guide.txt
>
>
> Cheers
> Nick
>
>

Re: Binutils Internals Document In PDF Format.

2009-12-21T02:53:37Z

Hi Prashant,

> I am new to Binutils and need the internals document ( Better if in
> 'PDF' format. ) so as to port Binutils to a new architecture.

A binutils contributor has recently made just such a document, although
it is in a plain text format rather than PDF. You can find it here:

http://sourceware.org/binutils/binutils-porting-guide.txt

Cheers
Nick

update my email

2009-12-21T02:20:22Z

bfd/
* elf64-ppc.c: Delete my email address.
binutils/
* MAINTAINERS: Update my email address.

Index: bfd/elf64-ppc.c
===================================================================
RCS file: /cvs/src/src/bfd/elf64-ppc.c,v
retrieving revision 1.305
diff -u -p -r1.305 elf64-ppc.c
--- bfd/elf64-ppc.c 17 Dec 2009 00:07:38 -0000 1.305
+++ bfd/elf64-ppc.c 21 Dec 2009 09:59:39 -0000
@@ -3,7 +3,7 @@
2009 Free Software Foundation, Inc.
Written by Linus Nordberg, Swox AB <info@...>,
based on elf32-ppc.c by Ian Lance Taylor.
- Largely rewritten by Alan Modra <amodra@...>
+ Largely rewritten by Alan Modra.

This file is part of BFD, the Binary File Descriptor library.

Index: binutils/MAINTAINERS
===================================================================
RCS file: /cvs/src/src/binutils/MAINTAINERS,v
retrieving revision 1.128
diff -u -p -r1.128 MAINTAINERS
--- binutils/MAINTAINERS 21 Oct 2009 03:10:53 -0000 1.128
+++ binutils/MAINTAINERS 21 Dec 2009 10:00:02 -0000
@@ -38,7 +38,7 @@ repository without obtaining approval fi
Jeff Law <law@...>
Jim Wilson <wilson@...>
DJ Delorie <dj@...>
- Alan Modra <amodra@...>
+ Alan Modra <amodra@...>
Michael Meissner <gnu@...>
Daniel Jacobowitz <dan@...>
Richard Sandiford <rdsandiford@...>
@@ -76,7 +76,7 @@ responsibility among the other maintaine
FRV Alexandre Oliva <aoliva@...>
H8300 Prafulla Thakare <prafulla.thakare@...>
HPPA Dave Anglin <dave.anglin@...>
- HPPA elf32 Alan Modra <amodra@...>
+ HPPA elf32 Alan Modra <amodra@...>
HPPA elf64 Jeff Law <law@...> [Basic maintainance only]
IA-64 Jim Wilson <wilson@...>
IQ2000 Stan Cox <scox@...>
@@ -102,7 +102,7 @@ responsibility among the other maintaine
MSP430 Dmitry Diky <diwil@...>
NetBSD support Matt Thomas <matt@...>
PPC Geoff Keating <geoffk@...>
- PPC Alan Modra <amodra@...>
+ PPC Alan Modra <amodra@...>
PPC vector ext Aldy Hernandez <aldyh@...>
RX DJ Delorie <dj@...>
RX Nick Clifton <nickc@...>
@@ -111,7 +111,7 @@ responsibility among the other maintaine
SH Alexandre Oliva <aoliva@...>
SH Kaz Kojima <kkojima@...>
SPARC Jakub Jelinek <jakub@...>
- SPU Alan Modra <amodra@...>
+ SPU Alan Modra <amodra@...>
TIC4X Svein Seldal <svein@...>
TIC54X Timothy Wall <twall@...>
VAX Matt Thomas <matt@...>

--
Alan Modra
Australia Development Lab, IBM

Re: Looking to contribute OMF support

2009-12-21T01:38:31Z

Bernd Jendrissek <berndj@...> wrote in
news:20060310140046.GA11152@...:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi all
>
> It seems my employer should be willing to assign copyright for the OMF
> support I've added using (in part) company time and equipment, to the
> FSF. I'll need one of those pre-paperwork-preparation forms that I've
> seen around here regularly; would somebody please send me one of
> those? Private email or here is fine.
>
> Since this will be certainly my, and likely the company's, first such
> submission, I'm a little daunted by the length (time) of the
> procedure. I need to fill in a form for requesting the real form,
> right? Is it only the second set of forms which have to cross the
> ocean by snail-mail?
>
> As for the OMF support itself:
>
> I wrote the object file format support against .OBJ files produced by
> TopSpeed C 3.10, which AFAICT is supposed to be compatible with the MS
> (and Borland too, probably) tools of the era, except that it defines a
> few comment records for specifying dependencies, and defines multiple
> sections with the same name (not sure how kosher that is to other
> OMF-aware tools).
>
> I didn't bother trying to write complete support, and I can't even
> guarantee that I interpret records correctly either; this is one of
> those "it works Well Enough For Me for what I want to do with it"
> ports.
>
> Notably, what works WEFM is:
> - section dumping and disassembly and of course 'size' which was my
> scratch to my itch of wanting to know which modules needed the most
> memory
> - listing of symbols and relocations (OMF relocs are hell, expect
> bugs) - surprisingly to me, ld was able correctly to locate a tiny
> dummy
> module that contains "offset16" relocs.
>
> NOT working or untested:
> - linking of more complex modules
> - objcopy
> - assembler output
>
> I hope that's good enough for an otherwise nonexistent port? (I'd
> prefer to show-and-tell once the paperwork is done.)
>
> - --
> You're proposing to build a box with a light on top of it. The light
> is supposed to go off when you carry the box into a room that has a
> Unicorn in it. How do you show that it works?
> - Dr. Gene "spaf" Spafford, at Dr. Wenliang Du's qualifing
> exam
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.2 (GNU/Linux)
> Comment: Please fetch my new key 804177F8 from
> hkp://wwwkeys.eu.pgp.net/
>
> iD8DBQFEEYZfwyMv24BBd/gRAnE7AJ48CJ7ceIk+aDn9kIY8FEx40wD7FACdG5hD
> dWN1uFxt4Lx42iTcJagxtTc=
> =Owyj
> -----END PGP SIGNATURE-----
>

This thread is kind of old but I was wondering whatever happened to this
endeavor of hacking OMF support into libbfd. Did it ever come to
fruitation?

I'm looking for borland/delphi source debugging support with gdb and
came across this thread. If OMF support is fully implemented into bfd,
would anything more have to be done to get source level debugging to
work in gdb for executables/dll's compiled with something other than
gcc? Specifically compiles that emit OMF symbolic debug info like
borland c++ build. I beleive Digital Mars Compiler emits OMF symbolic
info as well.

Thanks

Binutils Internals Document In PDF Format.

2009-12-20T21:09:30Z

Hi all,

I am new to Binutils and need the internals document ( Better if in
'PDF' format. ) so as to port Binutils to a new architecture.

Any help in this regard is appreciated.

Thanking in advance,
Prashant.

Re: Help about prelinker on mips

2009-12-20T18:22:42Z

Thanks very much.
After applied the two patches (both from ) in :
http://sourceware.org/ml/prelink/2008-q4/msg00001.html
the prelinker complained new messages like :
-----------------------
Assigned virtual address space slots for libraries:
/lib/ld.so.1 2c000000-2c02fa44
/lib/libc.so.6 2c040000-2c1a5b80
/lib/libdl.so.2 2c1b0000-2c1c20e0
/lib/libcrypt.so.1 2c1d0000-2c2101fc
/lib/libnsl.so.1 2c220000-2c2478b0
/lib/libresolv.so.2 2c250000-2c263960
/lib/libutil.so.1 2c270000-2c2820d0
Prelinking /mnt/nfs/tools/dummy
./prelink: /mnt/nfs/tools/dummy: Could not find symbol copy reloc is against

-----------------------
What does this mean? I found only one post about this message:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=411356

and it seems something wrong with the application, rather than tools.

Thanks.
--
Best Regards.

Re: R_MIPS_TLS_GD reloc overflow when linking >=xulrunner-1.9.1.5

2009-12-20T09:24:52Z

I made some progress, found where the problem happens:

(gdb) info b
Num Type Disp Enb Address What
2 breakpoint keep y 0x2ab390cc in mips_elf_calculate_relocation at elfxx-mips.c:5094
stop only if r_type==R_MIPS_TLS_GD
breakpoint already hit 2 times

return value:
5164 g = mips_elf_local_got_index (abfd, input_bfd, info,
Value returned is $17 = 73452

5171 g = mips_elf_got_offset_from_index (info, abfd, input_bfd, g);
Value returned is $18 = 40700

Then 40700 is passed to mips_elf_overflow_p(value, 16)

However, I haven't got time to take a look at what those two function does.
Will do it later.

BTW, the userland here is N32 ABI.

--
Zhang, Le
Gentoo/Loongson Developer
http://zhangle.is-a-geek.org
0260 C902 B8F8 6506 6586 2B90 BC51 C808 1E4E 2973

attachment0 (205 bytes) Download Attachment

[ob]: gas/xc16x/xc16x.exp (*): Add missing " in timeout cases.

2009-12-19T11:35:00Z

Hi.

I checked this in under the obvious rule.

2009-12-19 Doug Evans <dje@...>

* gas/xc16x/xc16x.exp (*): Add missing " in timeout cases.

Index: gas/xc16x/xc16x.exp
===================================================================
RCS file: /cvs/src/src/gas/testsuite/gas/xc16x/xc16x.exp,v
retrieving revision 1.2
diff -u -p -r1.2 xc16x.exp
--- gas/xc16x/xc16x.exp 19 Oct 2009 15:27:39 -0000 1.2
+++ gas/xc16x/xc16x.exp 19 Dec 2009 19:33:26 -0000
@@ -18,7 +18,7 @@ proc do_xc16x_add {} {
-re " +\[0-9\]+ 0008 06F0D204\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 000c 02F0EDFF\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0010 04F0EDFF\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -48,7 +48,7 @@ proc do_xc16x_addb {} {
-re " +\[0-9\]+ 0008 07F03300\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 000c 03F08723\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0010 05F08723\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -78,7 +78,7 @@ proc do_xc16x_addc {} {
-re " +\[0-9\]+ 000a 16F05634\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 000e 12F08723\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0012 14F08723\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -107,7 +107,7 @@ proc do_xc16x_addcb {} {
-re " +\[0-9\]+ 0006 17F00200\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 000a 17F02300\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 000e 15F08723\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -137,7 +137,7 @@ proc do_xc16x_sub {} {
-re " +\[0-9\]+ 0008 26F04376\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 000c 22F04376\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0010 24F04376\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -167,7 +167,7 @@ proc do_xc16x_subb {} {
-re " +\[0-9\]+ 0008 27F04300\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 000c 23F04376\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0010 25F04376\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -197,7 +197,7 @@ proc do_xc16x_subc {} {
-re " +\[0-9\]+ 0008 36F04300\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 000c 32F04376\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0010 34F04376\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -227,7 +227,7 @@ proc do_xc16x_subcb {} {
-re " +\[0-9\]+ 0008 37F04300\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 000c 33F04376\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0010 35F04376\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -257,7 +257,7 @@ proc do_xc16x_and {} {
-re " +\[0-9\]+ 0008 66F0BEFC\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 000c 62F03002\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0010 64F02003\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -287,7 +287,7 @@ proc do_xc16x_andb {} {
-re " +\[0-9\]+ 0008 67F0BE00\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 000c 63F03002\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0010 65F02003\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -317,7 +317,7 @@ proc do_xc16x_or {} {
-re " +\[0-9\]+ 0008 76F03402\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 000c 72F03645\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0010 74F03645\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -347,7 +347,7 @@ proc do_xc16x_xor {} {
-re " +\[0-9\]+ 0008 56F03402\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 000c 52F03402\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0010 54F03402\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -377,7 +377,7 @@ proc do_xc16x_xorb {} {
-re " +\[0-9\]+ 0008 57F03400\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 000c 53F00324\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0010 55F00324\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -417,7 +417,7 @@ proc do_xc16x_mov {} {
-re " +\[0-9\]+ 0024 9400CBFF\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0028 F2F0CBFF\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 002c F6F0CBFF\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -456,7 +456,7 @@ proc do_xc16x_movb {} {
-re " +\[0-9\]+ 0028 B4033412\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 002c F3F01200\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0030 F7F01200\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -482,7 +482,7 @@ proc do_xc16x_movbs {} {
-re " +\[0-9\]+ 0000 D020\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0002 D2F0FF00\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0006 D5F0CBFF\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -508,7 +508,7 @@ proc do_xc16x_movbz {} {
-re " +\[0-9\]+ 0000 C002\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0002 C2F0DD23\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0006 C5F02300\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -533,7 +533,7 @@ proc do_xc16x_pushpop {} {
expect {
-re " +\[0-9\]+ 0000 FCF0\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0002 ECF0\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -566,7 +566,7 @@ proc do_xc16x_shlrol {} {
-re " +\[0-9\]+ 000e 3C40\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0010 AC01\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0012 BC40\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -591,7 +591,7 @@ proc do_xc16x_neg {} {
expect {
-re " +\[0-9\]+ 0000 8100\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0002 A100\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -615,7 +615,7 @@ proc do_xc16x_mul {} {
expect {
-re " +\[0-9\]+ 0000 0B01\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0002 1B01\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -642,7 +642,7 @@ proc do_xc16x_div {} {
-re " +\[0-9\]+ 0002 6B00\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0004 7B00\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0006 5B00\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -684,7 +684,7 @@ proc do_xc16x_jmpa {} {
-re " +\[0-9\]+ 0040 EAD0AAAA\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0044 EAA0AAAA\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0048 EA10AAAA\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -726,7 +726,7 @@ proc do_xc16x_jmpi {} {
-re " +\[0-9\]+ 0020 9CD7\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0022 9CA7\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0024 9C17\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -770,7 +770,7 @@ proc do_xc16x_jmpr {} {
-re " +\[0-9\]+ 0022 ADEE\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0024 1DED\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0026 CDEC\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -801,7 +801,7 @@ proc do_xc16x_syscontrol1 {} {
-re " +\[0-9\]+ 0012 A55AA5A5\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0016 857A8585\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 001a B54AB5B5\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -840,7 +840,7 @@ proc do_xc16x_syscontrol2 {} {
-re " +\[0-9\]+ 001e D7300100\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0022 DCB5\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0024 D7B00100\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -865,7 +865,7 @@ proc do_xc16x_cpl {} {
expect {
-re " +\[0-9\]+ 0000 9100\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0002 B100\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -894,7 +894,7 @@ proc do_xc16x_cmp {} {
-re " +\[0-9\]+ 0006 4803\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0008 46F03402\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 000c 42F05234\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -923,7 +923,7 @@ proc do_xc16x_cmpb {} {
-re " +\[0-9\]+ 0006 4903\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0008 47F02200\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 000c 43F03402\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -958,7 +958,7 @@ proc do_xc16x_cmpi {} {
-re " +\[0-9\]+ 001e 90F0\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0020 96F0CBFC\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0024 92F0CBFF\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -1000,7 +1000,7 @@ proc do_xc16x_calli {} {
-re " +\[0-9\]+ 0020 ABD1\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0022 AB11\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0024 ABC1\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -1043,7 +1043,7 @@ proc do_xc16x_calla {} {
-re " +\[0-9\]+ 0044 CAA0AAAA\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0048 CA10AAAA\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 004c CAC0AAAA\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -1075,7 +1075,7 @@ proc do_xc16x_bit {} {
-re " +\[0-9\]+ 0010 5AF0F021\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0014 7AF0F021\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0018 2AF0F021\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -1100,7 +1100,7 @@ proc do_xc16x_bfldl {} {
expect {
-re " +\[0-9\]+ 0000 0AF0870E\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0004 1AF00EFF\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -1127,7 +1127,7 @@ proc do_xc16x_ret {} {
-re " +\[0-9\]+ 0002 FB88\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0004 DB00\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0006 EBF5\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -1151,7 +1151,7 @@ proc do_xc16x_trap {} {
while 1 {
expect {
-re " +\[0-9\]+ 0000 9B04\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -1181,7 +1181,7 @@ proc do_xc16x_orb {} {
-re " +\[0-9\]+ 0008 77F02300\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 000c 73F03402\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0010 75F03402\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -1206,7 +1206,7 @@ proc do_xc16x_prior {} {
while 1 {
expect {
-re " +\[0-9\]+ 0000 2B01\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -1231,7 +1231,7 @@ proc do_xc16x_nop {} {
expect {
-re " +\[0-9\]+ 0000 CC00\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0002 CC00\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}
@@ -1257,7 +1257,7 @@ proc do_xc16x_scxt {} {
expect {
-re " +\[0-9\]+ 0000 C6F0FFFF\[^\n\]*\n" { set x [expr $x+1] }
-re " +\[0-9\]+ 0004 D6F0FFFF\[^\n\]*\n" { set x [expr $x+1] }
- timeout { perror "timeout\n; break }
+ timeout { perror "timeout\n"; break }
eof { break }
}
}

PATCH: Replace VexNDS, VexNDD and VexLWP with VexVVVV.

2009-12-19T10:37:38Z

Hi,

I checked in this patch to peplace VexNDS, VexNDD and VexLWP with VexVVVV.

H.J.
----
gas/

2009-12-19 H.J. Lu <hongjiu.lu@...>

* config/tc-i386.c (process_operands): Check vexvvvv instead
of vexnds and vexndd.
(build_modrm_byte): Check vexvvvv instead of vexnds, vexndd
and vexlwp.

opcodes/

2009-12-19 H.J. Lu <hongjiu.lu@...>

* i386-gen.c (opcode_modifiers): Remove VexNDS, VexNDD and
VexLWP. Add VexVVVV.

* i386-opc.h (VexNDS): Removed.
(VexNDD): Likewise.
(VexLWP): Likewise.
(VEXXDS): New.
(VEXNDD): Likewise.
(VEXLWP): Likewise.
(VexVVVV): Likewise.
(i386_opcode_modifier): Remove vexnds, vexndd and vexlwp.
Add vexvvvv.

* i386-opc.tbl: Replace VexNDS with VexVVVV=1, VexNDD with
VexVVVV=2 and VexLWP with VexVVVV=3.
* i386-tbl.h: Regenerated.

x.bz2 (39K) Download Attachment

Re: Help about prelinker on mips

2009-12-19T09:37:33Z

I believe all binutils patches relevant to MIPS prelink are upstream. But
there are at least two patches discussed on the prelink list last December
that have not been reviewed or committed to the prelink repository.

http://sourceware.org/ml/prelink/2008-q4/msg00001.html
http://sourceware.org/ml/prelink/2008-q4/msg00006.html
Last ping: http://sourceware.org/ml/prelink/2009-q1/msg00000.html

It also looks like the older patch
http://sourceware.org/ml/prelink/2007-q4/msg00001.html
is not present either.

I have no idea whether any of these patches are relevant to your problem.

Note that none of the extensive cross-prelink patches are upstream.

--
Joseph S. Myers
joseph@...

Help about prelinker on mips

2009-12-19T03:19:24Z

Hi:
Don't know if this is the right list, Please correct me if not.
Recently I am trying to do prelink on mips, I am using
codesourcery released lite toolchain.
But it does not work(prelinker complained about conflicts and didn't
modify binaries), such as:

Assigned virtual address space slots for libraries:
/lib/ld.so.1 2c000000-2c02fa44
/lib/libc.so.6 2c040000-2c1a5b80
Prelinking /mnt/nfs/tools/dummy
./prelink: /lib/libc.so.6: Conflict 0000bce0 not found in any relocation
./prelink: /lib/libc.so.6: Conflict 0000bc20 not found in any relocation
./prelink: /lib/ld.so.1: Conflict 00000494 not found in any relocation
./prelink: /lib/ld.so.1: Conflict 00000484 not found in any relocation
./prelink: /lib/ld.so.1: Conflict 00000474 not found in any relocation
./prelink: /lib/ld.so.1: Conflict 00000464 not found in any relocation
./prelink: /lib/ld.so.1: Conflict 00000444 not found in any relocation

the version of tools are:
binutils-2.19
gcc-4.3.3
eglibc-2.8
elfutils-0.140
prelink-20091104(from http://people.redhat.com/jakub/prelink/)

I searched around www and found following patches and compared it
with source code:
ftp://ftp.linux-mips.org/pub/linux/mips/people/ths/mips-prelinker-patches-debian/

I found the glibc-patch are already merged, while the
binutils-patch is some kind of complicate:
most of binutils patches are applied, but ones like
binutils-objcopy-section-syms.patch not.
further more, the prelink patch are not applied, but the source code
do have a file named as "arch.mips.c",
which looks strange.

So, do I need any patch or have to do something special when
configuring/making/using
the toolchain/prelink ? Any tips will be highly appreciated.
Thanks All.

--
Best Regards.

Re: Adding -Wshadow to binutils builds

2009-12-19T00:34:28Z

Hi Steve,

> The -Wshadow change broke the gas build on hppa. Unless I hear an
> objection I will check in this change as an obvious fix. Tested on
> hppa2.0w-hp-hpux11.11.

No objections - please commit.

Cheers
Nick

> 2009-12-18 Steve Ellcey <sje@...>
>
> * config/tc-hppa.c: Change access to access_ctr.

Re: [PATCH] Objcopy: use a hash table for symbol renaming

2009-12-18T14:46:30Z

On Fri, 18 Dec 2009, Eirik Byrkjeflot Anonsen wrote:
> The attached patch replaces the linked list with a hash table
> implementation.

As just a random binutils target maintainer, I'd say: Thanks!

But... is there a good reason to add a new hash table
implementation instead of re-using the one in libiberty/hashtab.c?
(Besides, missing ChangeLog entry and required FSF assignment,
which I guess you'll hear more about from someone else.)

brgds, H-P

PATCH: Move Imm1 before Imm8

2009-12-18T13:05:48Z

I checked in this patch to move Imm1 before Imm8.

H.J.
---
diff --git a/opcodes/ChangeLog b/opcodes/ChangeLog
index 222603e..b5d9c00 100644
--- a/opcodes/ChangeLog
+++ b/opcodes/ChangeLog
@@ -1,3 +1,7 @@
+2009-12-18 H.J. Lu <hongjiu.lu@...>
+
+ * i386-gen.c (operand_types): Move Imm1 before Imm8.
+
2009-12-17 Nick Clifton <nickc@...>

PR binutils/10924
diff --git a/opcodes/i386-gen.c b/opcodes/i386-gen.c
index bc3f9e6..680b0b2 100644
--- a/opcodes/i386-gen.c
+++ b/opcodes/i386-gen.c
@@ -378,13 +378,13 @@ static bitfield operand_types[] =
BITFIELD (RegMMX),
BITFIELD (RegXMM),
BITFIELD (RegYMM),
+ BITFIELD (Imm1),
BITFIELD (Imm8),
BITFIELD (Imm8S),
BITFIELD (Imm16),
BITFIELD (Imm32),
BITFIELD (Imm32S),
BITFIELD (Imm64),
- BITFIELD (Imm1),
BITFIELD (BaseIndex),
BITFIELD (Disp8),
BITFIELD (Disp16),

Re: [RFA] Fix default endianness for gas mep.

2009-12-18T12:30:42Z

DJ Delorie wrote:
> gcc defaults mep to little endian too, I think the testsuite is wrong.
>

Joy. Ok, I'll submit an alternate patch that fixes the testsuite too
(probably by not making any assumptions of what the default is, e.g.
rename allinsn.d to allinsn.be.d, akin to the existing allinsn.le.d).

Re: [gold] "patch" for the incremental link design document

2009-12-18T12:29:39Z

I talked with Ian a bit more and he noticed that the second proposal
was incomplete. If, in the example of the previous email, a new
undefined reference shows up in a.o we wouldn't be able to find the
linked list of files for that symbol since the symbol was unused
before and therefore not in the global symbol table.

An updated proposal that I think addresses this issues is

--------------------------------
Change the "For an object file or shared library" section into:

For an object file, shared library or archive

1. 4 byte count of input sections

* Always zero for a shared library or archive.

2. 4 byte count of global symbols. In the case of an archive, this
will be the count of symbols in the *unused* members.

3. For each input section:

1. 4 byte offset to section name in .gnu_incremental_strtab section
2. 4 byte output section index
* If input section is omitted, this is zero.
3. 8 byte offset of input section in output section
4. 8 byte input section size

4. For each global symbol:

1. 4 byte index into output global symbol table if the symbol if an
object file or library. 4 byte offset into .gnu_incremental_strtab if
an archive.
2. 4 byte offset to next linked list entry
* Linked list starts from global symbol list.
* Offset is to entry in input file list.
3. 4 byte count of relocations for this file.
4. 4 byte offset to relocations in .gnu_incremental_relocs section.
---------------------------------

The change is that now t the symbols in unused archive members point
to the incremental string table. This should give us enough
information to build the necessary lists for each symbol.

For example, lets say b.o defines foo and is used. c.o defines bar and
is unused. Executing

ar q a,a b,o c,o
ld -incremental d.o a.a -o exec

Will produce (among others) the lists

foo -> b.o
bar -> a.a

now if we change d.o and add an undefined reference to bar we can just
look up the list for bar, find that a.a is on it and force a relink.

Cheers,
--
Rafael Ávila de Espíndola

Re: [RFA] Fix default endianness for gas mep.

2009-12-18T12:26:22Z

gcc defaults mep to little endian too, I think the testsuite is wrong.

Re: Adding -Wshadow to binutils builds

2009-12-18T08:37:47Z

The -Wshadow change broke the gas build on hppa. Unless I hear an
objection I will check in this change as an obvious fix. Tested on
hppa2.0w-hp-hpux11.11.

Steve Ellcey
sje@...

2009-12-18 Steve Ellcey <sje@...>

* config/tc-hppa.c: Change access to access_ctr.

Index: config/tc-hppa.c
===================================================================
RCS file: /cvs/src/src/gas/config/tc-hppa.c,v
retrieving revision 1.146
diff -r1.146 tc-hppa.c
7372c7372
< int i, access, space_index, alignment, quadrant, applicable, flags;
---
> int i, access_ctr, space_index, alignment, quadrant, applicable, flags;
7395c7395
< access = 0x7f;
---
> access_ctr = 0x7f;
7440c7440
< access = pa_def_subspaces[i].access;
---
> access_ctr = pa_def_subspaces[i].access;
7478c7478
< access = get_absolute_expression ();
---
> access_ctr = get_absolute_expression ();
7586c7586
< dup_common, sort, zero, access,
---
> dup_common, sort, zero, access_ctr,
7593c7593
< access, space_index,
---
> access_ctr, space_index,
7824c7824
< int access,
---
> int access_ctr,
7879c7879
< obj_set_subsection_attributes (seg, space->sd_seg, access, sort,
---
> obj_set_subsection_attributes (seg, space->sd_seg, access_ctr, sort,
7899c7899
< int access,
---
> int access_ctr,
7910c7910
< obj_set_subsection_attributes (section, space->sd_seg, access, sort,
---
> obj_set_subsection_attributes (section, space->sd_seg, access_ctr, sort,

Re: [PATCH] MIPS/GAS: Make .aent set function symbol type

2009-12-18T08:21:16Z

Hi Maciej,

> gas/
> * config/tc-mips.c (s_mips_ent): Also set BSF_FUNCTION for .aent.
>
> gas/testsuite/
> * gas/mips/aent.d: New test.
> * gas/mips/aent.s: Source for the new test.
> * gas/mips/mips.exp: Run it.

Approved - please apply.

Cheers
Nick

Re: [PATCH] MIPS/GAS: ERET test case failure

2009-12-18T08:20:02Z

Hi Maciej,

> 2009-12-14 Maciej W. Rozycki <macro@...>
>
> gas/testsuite/
> * gas/mips/eret-2.s: Add an instruction to fill a branch delay
> slot.
> * gas/mips/eret-2.d: Adjust accordingly.

Approved - please apply.

Cheers
Nick

Re: [patch, s390] Support NT_S390_HIGH_GPRS core file notes

2009-12-18T08:13:53Z

Hi Ulrich,

> include/elf/
> * common.h (NT_S390_HIGH_GPRS): Define.
>
> bfd/
> * elf.c (elfcore_grok_s390_high_gprs): New function.
> (elfcore_grok_note): Handle NT_S390_HIGH_GPRS notes.
> (elfcore_write_s390_high_gprs): New function.
> (elfcore_write_register_note): Call it.
>
> binutils/
> * readelf.c (get_note_type): Handle NT_S390_HIGH_GPRS notes.

Approved - please apply.

Cheers
Nick

[patch, s390] Support NT_S390_HIGH_GPRS core file notes

2009-12-18T07:36:45Z

Hello,

System z hardware allows use of the full 64-bit general purpose
register set (and corresponding instructions) even to 32-bit
applications running under a 64-bit kernel. Recent Linux kernels
have added support for this mode of operation. In particular,
this affects the core file format: when such an application
crashes, we need not just the usual 32-bit register set, but
the full 64-bit register contents in order to enable debugging.

To preserve compatibility, the kernel does not change the format
of the NT_PRSTATUS notes it generates; these continue to hold
the 32-bit register set, i.e. the lower halves of the actual
64-bit general registers. In addition, the kernel now generates
a new note section holding the *upper* halves of those registers.

This patch adds binutils support for this note section.

Tested on s390x-linux; also tested as part of a GDB tree in
conjunction with an (upcoming) GDB patch to add support for
debugging such applications.

OK for mainline?

Bye,
Ulrich

ChangeLog:

include/elf/
* common.h (NT_S390_HIGH_GPRS): Define.

bfd/
* elf.c (elfcore_grok_s390_high_gprs): New function.
(elfcore_grok_note): Handle NT_S390_HIGH_GPRS notes.
(elfcore_write_s390_high_gprs): New function.
(elfcore_write_register_note): Call it.

binutils/
* readelf.c (get_note_type): Handle NT_S390_HIGH_GPRS notes.

Index: bfd/elf.c
===================================================================
RCS file: /cvs/src/src/bfd/elf.c,v
retrieving revision 1.495
diff -u -p -r1.495 elf.c
--- bfd/elf.c 11 Dec 2009 13:42:02 -0000 1.495
+++ bfd/elf.c 18 Dec 2009 15:23:49 -0000
@@ -7638,6 +7638,12 @@ elfcore_grok_ppc_vsx (bfd *abfd, Elf_Int
return elfcore_make_note_pseudosection (abfd, ".reg-ppc-vsx", note);
}

+static bfd_boolean
+elfcore_grok_s390_high_gprs (bfd *abfd, Elf_Internal_Note *note)
+{
+ return elfcore_make_note_pseudosection (abfd, ".reg-s390-high-gprs", note);
+}
+
#if defined (HAVE_PRPSINFO_T)
typedef prpsinfo_t elfcore_psinfo_t;
#if defined (HAVE_PRPSINFO32_T) /* Sparc64 cross Sparc32 */
@@ -7999,6 +8005,13 @@ elfcore_grok_note (bfd *abfd, Elf_Intern
else
return TRUE;

+ case NT_S390_HIGH_GPRS:
+ if (note->namesz == 6
+ && strcmp (note->namedata, "LINUX") == 0)
+ return elfcore_grok_s390_high_gprs (abfd, note);
+ else
+ return TRUE;
+
case NT_PRPSINFO:
case NT_PSINFO:
if (bed->elf_backend_grok_psinfo)
@@ -8641,6 +8654,19 @@ elfcore_write_ppc_vsx (bfd *abfd,
note_name, NT_PPC_VSX, ppc_vsx, size);
}

+static char *
+elfcore_write_s390_high_gprs (bfd *abfd,
+ char *buf,
+ int *bufsiz,
+ const void *s390_high_gprs,
+ int size)
+{
+ char *note_name = "LINUX";
+ return elfcore_write_note (abfd, buf, bufsiz,
+ note_name, NT_S390_HIGH_GPRS,
+ s390_high_gprs, size);
+}
+
char *
elfcore_write_register_note (bfd *abfd,
char *buf,
@@ -8657,6 +8683,8 @@ elfcore_write_register_note (bfd *abfd,
return elfcore_write_ppc_vmx (abfd, buf, bufsiz, data, size);
if (strcmp (section, ".reg-ppc-vsx") == 0)
return elfcore_write_ppc_vsx (abfd, buf, bufsiz, data, size);
+ if (strcmp (section, ".reg-s390-high-gprs") == 0)
+ return elfcore_write_s390_high_gprs (abfd, buf, bufsiz, data, size);
return NULL;
}

Index: binutils/readelf.c
===================================================================
RCS file: /cvs/src/src/binutils/readelf.c,v
retrieving revision 1.475
diff -u -p -r1.475 readelf.c
--- binutils/readelf.c 11 Dec 2009 13:42:06 -0000 1.475
+++ binutils/readelf.c 18 Dec 2009 15:23:51 -0000
@@ -10385,6 +10385,8 @@ get_note_type (unsigned e_type)
return _("NT_PPC_VMX (ppc Altivec registers)");
case NT_PPC_VSX:
return _("NT_PPC_VSX (ppc VSX registers)");
+ case NT_S390_HIGH_GPRS:
+ return _("NT_S390_HIGH_GPRS (s390 upper register halves)");
case NT_PSTATUS:
return _("NT_PSTATUS (pstatus structure)");
case NT_FPREGS:
Index: include/elf/common.h
===================================================================
RCS file: /cvs/src/src/include/elf/common.h,v
retrieving revision 1.117
diff -u -p -r1.117 common.h
--- include/elf/common.h 28 Nov 2009 03:18:00 -0000 1.117
+++ include/elf/common.h 18 Dec 2009 15:23:53 -0000
@@ -510,6 +510,8 @@
/* note name must be "LINUX". */
#define NT_PPC_VSX 0x102 /* PowerPC VSX registers */
/* note name must be "LINUX". */
+#define NT_S390_HIGH_GPRS 0x300 /* S/390 upper halves of GPRs */
+ /* note name must be "LINUX". */

/* Note segments for core files on dir-style procfs systems. */

--
Dr. Ulrich Weigand
GNU Toolchain for Linux on System z and Cell BE
Ulrich.Weigand@...

Re: [PATCH] Buffer overrun in objcopy

2009-12-18T06:37:24Z

Tristan Gingold <gingold@...> writes:

> On Dec 18, 2009, at 3:08 PM, Eirik Byrkjeflot Anonsen wrote:
>
>> Using objcopy from binutils 2.20.
>>
>> When using objcopy to rename symbols using add_redefine_syms_file(), if
>> one of the source symbols is exactly 99 characters, the input buffer
>> will be overrun (by the first character in the target symbol). The
>> attached patch copies the buffer resize code to the two places in this
>> function where I think it could potentially be a problem.
>
> Good catch. However I think it would be simpler to just allocate bufsize + 1 bytes.
>
> (I also think that this should go into the branch).
>
> Tristan.

Yes, that should work just fine. It might be slightly less obvious to
someone reading the code later, but I'm happy with either solution.

eirik

[PATCH] Objcopy: use a hash table for symbol renaming

2009-12-18T06:28:27Z

Using objcopy from binutils 2.20:

Symbol renaming in objcopy (from binutils 2.20) uses a linked list and a
linear search through that list. This is a tad slow when the symbol
list is long (in my case with ~45000 symbols it takes around 20
minutes).

The attached patch replaces the linked list with a hash table
implementation. I have taken care to retain the behaviour of the old
code (in particular, failing with an error message when the same symbol
appears twice, either as source or target symbol.) This makes the case
above complete in 5 seconds. And it even seems to do the right thing :)

Obvious points of possible improvements:

- A better hash function (sym_hash()).
- A better choice of hash table sizes.
- A better algorithm for when to resize the hash table.

Note: I'm off for christmas vacation now, so I may not be responding to
replies to this mail until next year.

eirik

From f5fbb2bc0d49c36e6b7c694e7112e621dcaca314 Mon Sep 17 00:00:00 2001 From: Eirik Byrkjeflot Anonsen <eirik@...> Date: Fri, 18 Dec 2009 14:33:38 +0100 Subject: [PATCH 1/1] Replace symbol redefinition list with a hash table. --- binutils/objcopy.c | 150 ++++++++++++++++++++++++++++++++++++++++++++++------ 1 files changed, 134 insertions(+), 16 deletions(-) diff --git a/binutils/objcopy.c b/binutils/objcopy.c index f92bdca..0718948 100644 --- a/binutils/objcopy.c +++ b/binutils/objcopy.c @@ -57,12 +57,19 @@ struct is_specified_symbol_predicate_data bfd_boolean found; }; -/* A list to support redefine_sym. */ +/* An entry in the collection of symbols to rename. + * + * These entries are stored in two hash tables using open hashing. + * One hash table is hashed on 'source', and the other on 'target'. + * 'next_source' and 'next_target' are the linked lists in the two + * hash tables. + */ struct redefine_node { char *source; char *target; - struct redefine_node *next; + struct redefine_node *next_source; + struct redefine_node *next_target; }; typedef struct section_rename @@ -214,7 +221,10 @@ static htab_t localize_specific_htab = NULL; static htab_t globalize_specific_htab = NULL; static htab_t keepglobal_specific_htab = NULL; static htab_t weaken_specific_htab = NULL; -static struct redefine_node *redefine_sym_list = NULL; +static struct redefine_node **redefine_sym_hash = NULL; +static struct redefine_node **redefine_sym_hash_target = NULL; +static size_t redefine_sym_hash_size = 0; +static size_t redefine_sym_entry_count = 0; /* If this is TRUE, we weaken global symbols (set BSF_WEAK). */ static bfd_boolean weaken = FALSE; @@ -1013,7 +1023,7 @@ filter_symbols (bfd *abfd, bfd *obfd, asymbol **osyms, undefined = bfd_is_und_section (bfd_get_section (sym)); - if (redefine_sym_list) + if (redefine_sym_hash) { char *old_name, *new_name; @@ -1175,14 +1185,91 @@ filter_symbols (bfd *abfd, bfd *obfd, asymbol **osyms, return dst_count; } +/* redefine_sym_hash helper functions. */ + +static size_t +sym_hash(const char * sym) +{ + /* Use sdbm hashing, for no particular reason. */ + size_t hash; + hash = 0; + while (*sym != 0) + { + hash = *sym + (hash << 6) + (hash << 16) - hash; + sym ++; + }; + return hash; +} + +/* Grow the hash tables. */ +static void +reallocate_redefine_sym_hash(void) +{ + size_t i; + size_t new_size; + struct redefine_node ** new_hash; + struct redefine_node ** new_hash_target; + /* Simple choice of new size. Can probably be improved. */ + new_size = redefine_sym_hash_size * 3 + 1; + if (new_size < 241) + new_size = 241; + else if (new_size > redefine_sym_entry_count) + /* Avoid making too big hash tables. It's better to have longer lists. */ + return; + + new_hash = (struct redefine_node **) xmalloc(sizeof(struct redefine_node) * new_size); + new_hash_target = (struct redefine_node **) xmalloc(sizeof(struct redefine_node) * new_size); + for (i = 0; i < new_size; i++) + { + new_hash[i] = 0; + new_hash_target[i] = 0; + } + + for (i = 0; i < redefine_sym_hash_size; i++) + { + while (redefine_sym_hash[i] != NULL) + { + struct redefine_node * sym; + size_t source_hash; + sym = redefine_sym_hash[i]; + redefine_sym_hash[i] = sym->next_source; + source_hash = sym_hash(sym->source); + sym->next_source = new_hash[source_hash % new_size]; + new_hash[source_hash % new_size] = sym; + }; + } + + for (i = 0; i < redefine_sym_hash_size; i++) + { + while (redefine_sym_hash_target[i] != NULL) + { + struct redefine_node * sym; + size_t target_hash; + sym = redefine_sym_hash_target[i]; + redefine_sym_hash_target[i] = sym->next_target; + target_hash = sym_hash(sym->target); + sym->next_target = new_hash_target[target_hash % new_size]; + new_hash_target[target_hash % new_size] = sym; + }; + } + + free(redefine_sym_hash); + free(redefine_sym_hash_target); + redefine_sym_hash = new_hash; + redefine_sym_hash_target = new_hash_target; + redefine_sym_hash_size = new_size; +}; + /* Find the redefined name of symbol SOURCE. */ static const char * lookup_sym_redefinition (const char *source) { struct redefine_node *list; + size_t hash_entry; + hash_entry = sym_hash(source) % redefine_sym_hash_size; - for (list = redefine_sym_list; list != NULL; list = list->next) + for (list = redefine_sym_hash[hash_entry]; list != NULL; list = list->next_source) if (strcmp (source, list->source) == 0) return list->target; @@ -1194,28 +1281,59 @@ lookup_sym_redefinition (const char *source) static void redefine_list_append (const char *cause, const char *source, const char *target) { - struct redefine_node **p; - struct redefine_node *list; - struct redefine_node *new_node; + size_t source_hash; + size_t target_hash; + size_t chain_depth; + size_t max_chain_depth; + struct redefine_node ** list; + struct redefine_node * new_node; - for (p = &redefine_sym_list; (list = *p) != NULL; p = &list->next) + /* If any list is longer than this, try to resize the hash table. */ + max_chain_depth = 10; + + if (redefine_sym_hash_size == 0) + { + reallocate_redefine_sym_hash(); + if (redefine_sym_hash_size == 0) + fatal(_("%s: Failed to allocate memory for list of redefined symbols"), + cause); + } + + source_hash = sym_hash(source); + target_hash = sym_hash(target); + chain_depth = 0; + for (list = &redefine_sym_hash[source_hash % redefine_sym_hash_size]; + *list != NULL; + list = &((*list)->next_source)) { - if (strcmp (source, list->source) == 0) + chain_depth ++; + if (strcmp (source, (*list)->source) == 0) fatal (_("%s: Multiple redefinition of symbol \"%s\""), cause, source); - - if (strcmp (target, list->target) == 0) + } + if (chain_depth <= max_chain_depth) + chain_depth = 0; + for (list = &redefine_sym_hash_target[target_hash % redefine_sym_hash_size]; + *list != NULL; + list = &((*list)->next_target)) + { + chain_depth ++; + if (strcmp (target, (*list)->target) == 0) fatal (_("%s: Symbol \"%s\" is target of more than one redefinition"), cause, target); } + if (chain_depth > max_chain_depth) + reallocate_redefine_sym_hash(); + redefine_sym_entry_count ++; new_node = (struct redefine_node *) xmalloc (sizeof (struct redefine_node)); new_node->source = strdup (source); new_node->target = strdup (target); - new_node->next = NULL; - - *p = new_node; + new_node->next_source = redefine_sym_hash[source_hash % redefine_sym_hash_size]; + new_node->next_target = redefine_sym_hash_target[target_hash % redefine_sym_hash_size]; + redefine_sym_hash[source_hash % redefine_sym_hash_size] = new_node; + redefine_sym_hash_target[target_hash % redefine_sym_hash_size] = new_node; } /* Handle the --redefine-syms option. Read lines containing "old new" @@ -1815,7 +1933,7 @@ copy_object (bfd *ibfd, bfd *obfd) || convert_debugging || change_leading_char || remove_leading_char - || redefine_sym_list + || redefine_sym_hash || weaken) { /* Mark symbols used in output relocations so that they -- 1.6.3.3

Re: [PATCH] Buffer overrun in objcopy

2009-12-18T06:15:29Z

On Dec 18, 2009, at 3:08 PM, Eirik Byrkjeflot Anonsen wrote:

> Using objcopy from binutils 2.20.
>
> When using objcopy to rename symbols using add_redefine_syms_file(), if
> one of the source symbols is exactly 99 characters, the input buffer
> will be overrun (by the first character in the target symbol). The
> attached patch copies the buffer resize code to the two places in this
> function where I think it could potentially be a problem.

Good catch. However I think it would be simpler to just allocate bufsize + 1 bytes.

(I also think that this should go into the branch).

Tristan.

[PATCH] Buffer overrun in objcopy

2009-12-18T06:08:27Z

Using objcopy from binutils 2.20.

When using objcopy to rename symbols using add_redefine_syms_file(), if
one of the source symbols is exactly 99 characters, the input buffer
will be overrun (by the first character in the target symbol). The
attached patch copies the buffer resize code to the two places in this
function where I think it could potentially be a problem.

eirik

From c62580d435fd5052c3b1173e94545ea12bcab8e9 Mon Sep 17 00:00:00 2001 From: Eirik Byrkjeflot Anonsen <eirik@...> Date: Fri, 18 Dec 2009 13:54:44 +0100 Subject: [PATCH 1/2] Extend read buffer to avoid buffer overruns. --- binutils/objcopy.c | 10 ++++++++++ 1 files changed, 10 insertions(+), 0 deletions(-) diff --git a/binutils/objcopy.c b/binutils/objcopy.c index 2048827..f92bdca 100644 --- a/binutils/objcopy.c +++ b/binutils/objcopy.c @@ -1259,6 +1259,11 @@ add_redefine_syms_file (const char *filename) c = getc (file); } buf[len++] = '\0'; + if (len >= bufsize) + { + bufsize *= 2; + buf = (char *) xrealloc (buf, bufsize); + } if (c == EOF) break; @@ -1285,6 +1290,11 @@ add_redefine_syms_file (const char *filename) c = getc (file); } buf[len++] = '\0'; + if (len >= bufsize) + { + bufsize *= 2; + buf = (char *) xrealloc (buf, bufsize); + } if (c == EOF) break; -- 1.6.3.3