Fix arm-linux bootstrap problem on cvs.

Fix arm-linux bootstrap problem on cvs.

[jeroen dobbelaere]

[-- Attachment #1: Type: text/plain, Size: 328 bytes --]

Attached patch fixes the bootstrapping problem of gcc (main cvs)
on the arm-linux target (with --enable-checking).

Currently bootstrapping (and this time without a problem
reaching stage1).

Is it OK to commit ?

Greetings,
-- 
Jeroen Dobbelaere
Embedded Software Engineer

ACUNIA Embedded Solutions
http://www.acunia.com/aes


[-- Attachment #2: gcc-head-arm-linux-build-jdo-1.patch --]
[-- Type: text/plain, Size: 2659 bytes --]

Index: gcc/gcc/ChangeLog
===================================================================
RCS file: /cvs/gcc/gcc/gcc/ChangeLog,v
retrieving revision 1.14773
diff -u -r1.14773 ChangeLog
--- gcc/gcc/ChangeLog	3 Jul 2002 09:49:38 -0000	1.14773
+++ gcc/gcc/ChangeLog	3 Jul 2002 11:49:56 -0000
@@ -1,3 +1,15 @@
+2002-07-03  Jeroen Dobbelaere <jeroen.dobbelaere@acunia.com>
+
+	* config/arm/arm.c (emit_sfm): Omit RTX_FRAME_RELATED_P(dwarf) = 1
+	as this breaks the 'enable-checking'. The dwarf expression itself
+	is only used inside a 'RTX_FRAME_RELATED_EXPR' note.
+
+2002-07-03  Jeroen Dobbelaere <jeroen.dobbelaere@acunia.com>
+
+	* config/arm/arm.h (LEGITIMATE_PIC_OPERAND_P): Only use 
+	CONSTANT_POOL_ADDRESS_P on SYMBOL_REF's. Rearrange the
+	resulting expression following Richard Earnshaw's proposal.
+
 Wed Jul  3 10:24:16 2002  J"orn Rennecke <joern.rennecke@superh.com>
 
 	* optabs.c (expand_vector_binop): Don't store using a SUBREG smaller
Index: gcc/gcc/config/arm/arm.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/config/arm/arm.c,v
retrieving revision 1.214
diff -u -r1.214 arm.c
--- gcc/gcc/config/arm/arm.c	1 Jul 2002 23:07:06 -0000	1.214
+++ gcc/gcc/config/arm/arm.c	3 Jul 2002 11:49:58 -0000
@@ -7922,7 +7922,6 @@
 
   par = gen_rtx_PARALLEL (VOIDmode, rtvec_alloc (count));
   dwarf = gen_rtx_PARALLEL (VOIDmode, rtvec_alloc (count));
-  RTX_FRAME_RELATED_P (dwarf) = 1;
 
   reg = gen_rtx_REG (XFmode, base_reg++);
 
Index: gcc/gcc/config/arm/arm.h
===================================================================
RCS file: /cvs/gcc/gcc/gcc/config/arm/arm.h,v
retrieving revision 1.152
diff -u -r1.152 arm.h
--- gcc/gcc/config/arm/arm.h	10 Jun 2002 22:35:45 -0000	1.152
+++ gcc/gcc/config/arm/arm.h	3 Jul 2002 11:49:58 -0000
@@ -2397,12 +2397,13 @@
 /* We can't directly access anything that contains a symbol,
    nor can we indirect via the constant pool.  */
 #define LEGITIMATE_PIC_OPERAND_P(X)					\
-	(   ! symbol_mentioned_p (X)					\
-	 && ! label_mentioned_p (X)					\
-	 && (! CONSTANT_POOL_ADDRESS_P (X)				\
-	     || (   ! symbol_mentioned_p (get_pool_constant (X))  	\
-	         && ! label_mentioned_p (get_pool_constant (X)))))
-     
+	(!(symbol_mentioned_p (X)					\
+	   || label_mentioned_p (X)					\
+	   || (GET_CODE (X) == SYMBOL_REF				\
+	       && CONSTANT_POOL_ADDRESS_P (X)				\
+	       && (symbol_mentioned_p (get_pool_constant (X))		\
+		   || label_mentioned_p (get_pool_constant (X))))))
+
 /* We need to know when we are making a constant pool; this determines
    whether data needs to be in the GOT or can be referenced via a GOT
    offset.  */

Re: Fix arm-linux bootstrap problem on cvs.

[Richard Earnshaw]

> Attached patch fixes the bootstrapping problem of gcc (main cvs)
> on the arm-linux target (with --enable-checking).
> 
> Currently bootstrapping (and this time without a problem
> reaching stage1).
> 
> Is it OK to commit ?
> 

Ok once bootstrap and regtest are successfully completed.  However, 
ChangeLog entries should be:


> +	* config/arm/arm.c (emit_sfm): Omit RTX_FRAME_RELATED_P(dwarf) = 1
> +	as this breaks the 'enable-checking'. The dwarf expression itself
> +	is only used inside a 'RTX_FRAME_RELATED_EXPR' note.

	* arm.c (emit_sfm): Don't set RTX_FRAME_RELATED_P on DWARF.

> +
> +
> +	* config/arm/arm.h (LEGITIMATE_PIC_OPERAND_P): Only use 
> +	CONSTANT_POOL_ADDRESS_P on SYMBOL_REF's. Rearrange the
> +	resulting expression following Richard Earnshaw's proposal.
> +

2002-07-03  Jeroen Dobbelaere <jeroen.dobbelaere@acunia.com>
	    Richard Earnshaw  <rearnsha@arm.com>

	* arm.h (LEGITIMATE_PIC_OPERAND_P): Only test 
	CONSTANT_POOL_ADDRESS_P on a SYMBOL_REF.  Simplify logic.


R.

Re: Fix arm-linux bootstrap problem on cvs.

[jeroen dobbelaere]

Richard Earnshaw wrote:
>>Attached patch fixes the bootstrapping problem of gcc (main cvs)
>>on the arm-linux target (with --enable-checking).
>>
>>Currently bootstrapping (and this time without a problem
>>reaching stage1).
>>
>>Is it OK to commit ?
>>
> 
> 
> Ok once bootstrap and regtest are successfully completed.  However, 
> ChangeLog entries should be:
> 
[..]

This will be difficult : for the last few weeks, the tests could not be run
as the compiler didn't work.

The bootstrap for just c,c++ worked fine.

I'm currently bootstrapping with all languages enabled (except for ada).
I'll do a 'make -k check' overnight. Results can be expected tomorrow morning (CET).

Greetings,
-- 
Jeroen Dobbelaere
Embedded Software Engineer

ACUNIA Embedded Solutions
http://www.acunia.com/aes

Re: Fix arm-linux bootstrap problem on cvs.

[Richard Earnshaw]

> > 
> > Ok once bootstrap and regtest are successfully completed.  However, 
> > ChangeLog entries should be:
> > 
> [..]
> 
> This will be difficult : for the last few weeks, the tests could not be run
> as the compiler didn't work.
> 

It should be possible to compare it with an unmodified compiler built with 
checking disabled.

R.

Re: Fix arm-linux bootstrap problem on cvs.

[jeroen dobbelaere]

Richard Earnshaw wrote:
>>>Ok once bootstrap and regtest are successfully completed.  However, 
>>>ChangeLog entries should be:
>>>
>>
>>[..]
>>
>>This will be difficult : for the last few weeks, the tests could not be run
>>as the compiler didn't work.
>>
> 
> 
> It should be possible to compare it with an unmodified compiler built with 
> checking disabled.
> 
> R.
> 

Ok. I'll do that. I'll try to start this up this evening...

Greetings,
-- 
Jeroen Dobbelaere
Embedded Software Engineer

ACUNIA Embedded Solutions
http://www.acunia.com/aes

[patch] Re: Fix arm-linux bootstrap problem on cvs.

[jeroen dobbelaere]

[-- Attachment #1: Type: text/plain, Size: 3352 bytes --]

Richard Earnshaw wrote:
>>>Ok once bootstrap and regtest are successfully completed.  However, 
>>>ChangeLog entries should be:
>>>
>>
>>[..]
>>
>>This will be difficult : for the last few weeks, the tests could not be run
>>as the compiler didn't work.
>>
> 
> 
> It should be possible to compare it with an unmodified compiler built with 
> checking disabled.
> 
> R.
> 

I've made a bootstrap build with tests for the cvs + patch
See <http://gcc.gnu.org/ml/gcc-testresults/2002-07/msg00081.html>

Bootstrapping the cvs without this patch (with --disable-checking) failed :


/usr/src/cvs-others/gcc-trees/gcc-head/build-cvs/gcc/xgcc -shared-libgcc -B/usr/src/cvs-others/gcc-trees/gcc-head/build-cvs/gcc/
  -nostdinc++ -L/usr/src/cvs-others/gcc-trees/gcc-head/build-cvs/armv5l-unknown-linux-gnu/libstdc++-v3/src -L/usr/src/cvs-others/
gcc-trees/gcc-head/build-cvs/armv5l-unknown-linux-gnu/libstdc++-v3/src/.libs -B/usr/local/gcc-3.2-cvs/armv5l-unknown-linux-gnu/b
in/ -B/usr/local/gcc-3.2-cvs/armv5l-unknown-linux-gnu/lib/ -isystem /usr/local/gcc-3.2-cvs/armv5l-unknown-linux-gnu/include -nos
tdinc++ -I/usr/src/cvs-others/gcc-trees/gcc-head/build-cvs/armv5l-unknown-linux-gnu/libstdc++-v3/include/armv5l-unknown-linux-gn
u -I/usr/src/cvs-others/gcc-trees/gcc-head/build-cvs/armv5l-unknown-linux-gnu/libstdc++-v3/include -I../../../../gcc/libstdc++-v
3/libsupc++ -I../../../../gcc/libstdc++-v3/libmath -g -O2 -D_GNU_SOURCE -fno-implicit-templates -Wall -Wno-format -W -Wwrite-str
ings -Winline -fdiagnostics-show-location=once -ffunction-sections -fdata-sections -g -c ../../../../gcc/libstdc++-v3/src/comple
x_io.cc  -fPIC -DPIC -o .libs/complex_io.o
In file included from /usr/src/cvs-others/gcc-trees/gcc-head/build-cvs/armv5l-unknown-linux-gnu/libstdc++-v3/include/ios:49,
                  from /usr/src/cvs-others/gcc-trees/gcc-head/build-cvs/armv5l-unknown-linux-gnu/libstdc++-v3/include/istream:44,
                  from /usr/src/cvs-others/gcc-trees/gcc-head/build-cvs/armv5l-unknown-linux-gnu/libstdc++-v3/include/sstream:44,
                  from /usr/src/cvs-others/gcc-trees/gcc-head/build-cvs/armv5l-unknown-linux-gnu/libstdc++-v3/include/complex:51,
                  from ../../../../gcc/libstdc++-v3/src/complex_io.cc:30:
/usr/src/cvs-others/gcc-trees/gcc-head/build-cvs/armv5l-unknown-linux-gnu/libstdc++-v3/include/bits/ios_base.h:427: internal com
piler error: internal
    error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://www.gnu.org/software/gcc/bugs.html> for instructions.
make[4]: *** [complex_io.lo] Error 1
make[4]: Leaving directory `/usr/src/cvs-others/gcc-trees/gcc-head/build-cvs/armv5l-unknown-linux-gnu/libstdc++-v3/src'
make[3]: *** [all-recursive] Error 1
make[3]: Leaving directory `/usr/src/cvs-others/gcc-trees/gcc-head/build-cvs/armv5l-unknown-linux-gnu/libstdc++-v3'
make[2]: *** [all-recursive-am] Error 2
make[2]: Leaving directory `/usr/src/cvs-others/gcc-trees/gcc-head/build-cvs/armv5l-unknown-linux-gnu/libstdc++-v3'
make[1]: *** [all-target-libstdc++-v3] Error 2
make[1]: Leaving directory `/usr/src/cvs-others/gcc-trees/gcc-head/build-cvs'
make: *** [bootstrap] Error 2

Attached is the patch, with adapted changelog.

Greetings,
-- 
Jeroen Dobbelaere
Embedded Software Engineer

ACUNIA Embedded Solutions
http://www.acunia.com/aes


[-- Attachment #2: gcc-head-arm-linux-build-jdo-2.patch --]
[-- Type: text/plain, Size: 2503 bytes --]

Index: gcc/gcc/ChangeLog
===================================================================
RCS file: /cvs/gcc/gcc/gcc/ChangeLog,v
retrieving revision 1.14773
diff -u -r1.14773 ChangeLog
--- gcc/gcc/ChangeLog	3 Jul 2002 09:49:38 -0000	1.14773
+++ gcc/gcc/ChangeLog	3 Jul 2002 11:49:56 -0000
@@ -1,3 +1,13 @@
+2002-07-03  Jeroen Dobbelaere <jeroen.dobbelaere@acunia.com>
+
+	* arm.c (emit_sfm): Don't set RTX_FRAME_RELATED_P on DWARF.
+
+2002-07-03  Jeroen Dobbelaere <jeroen.dobbelaere@acunia.com>
+	    Richard Earnshaw  <rearnsha@arm.com>
+
+	* arm.h (LEGITIMATE_PIC_OPERAND_P): Only test 
+	CONSTANT_POOL_ADDRESS_P on a SYMBOL_REF.  Simplify logic.
+
 Wed Jul  3 10:24:16 2002  J"orn Rennecke <joern.rennecke@superh.com>
 
 	* optabs.c (expand_vector_binop): Don't store using a SUBREG smaller
Index: gcc/gcc/config/arm/arm.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/config/arm/arm.c,v
retrieving revision 1.214
diff -u -r1.214 arm.c
--- gcc/gcc/config/arm/arm.c	1 Jul 2002 23:07:06 -0000	1.214
+++ gcc/gcc/config/arm/arm.c	3 Jul 2002 11:49:58 -0000
@@ -7922,7 +7922,6 @@
 
   par = gen_rtx_PARALLEL (VOIDmode, rtvec_alloc (count));
   dwarf = gen_rtx_PARALLEL (VOIDmode, rtvec_alloc (count));
-  RTX_FRAME_RELATED_P (dwarf) = 1;
 
   reg = gen_rtx_REG (XFmode, base_reg++);
 
Index: gcc/gcc/config/arm/arm.h
===================================================================
RCS file: /cvs/gcc/gcc/gcc/config/arm/arm.h,v
retrieving revision 1.152
diff -u -r1.152 arm.h
--- gcc/gcc/config/arm/arm.h	10 Jun 2002 22:35:45 -0000	1.152
+++ gcc/gcc/config/arm/arm.h	3 Jul 2002 11:49:58 -0000
@@ -2397,12 +2397,13 @@
 /* We can't directly access anything that contains a symbol,
    nor can we indirect via the constant pool.  */
 #define LEGITIMATE_PIC_OPERAND_P(X)					\
-	(   ! symbol_mentioned_p (X)					\
-	 && ! label_mentioned_p (X)					\
-	 && (! CONSTANT_POOL_ADDRESS_P (X)				\
-	     || (   ! symbol_mentioned_p (get_pool_constant (X))  	\
-	         && ! label_mentioned_p (get_pool_constant (X)))))
-     
+	(!(symbol_mentioned_p (X)					\
+	   || label_mentioned_p (X)					\
+	   || (GET_CODE (X) == SYMBOL_REF				\
+	       && CONSTANT_POOL_ADDRESS_P (X)				\
+	       && (symbol_mentioned_p (get_pool_constant (X))		\
+		   || label_mentioned_p (get_pool_constant (X))))))
+
 /* We need to know when we are making a constant pool; this determines
    whether data needs to be in the GOT or can be referenced via a GOT
    offset.  */

Re: [patch] Re: Fix arm-linux bootstrap problem on cvs.

[Richard Earnshaw]

jeroen.dobbelaere@acunia.com said:
> Bootstrapping the cvs without this patch (with --disable-checking)
> failed :


Can't reproduce this on my netwinder.  It also seems very strange that you 
should be seeing this on a --disable-checking build but not if you apply 
the patch and then enable checking.

Can you get a back-trace?

R.

Re: [patch] Re: Fix arm-linux bootstrap problem on cvs.

[jeroen dobbelaere]

Richard Earnshaw wrote:
> jeroen.dobbelaere@acunia.com said:
> 
>>Bootstrapping the cvs without this patch (with --disable-checking)
>>failed :
> 
> 
> 
> Can't reproduce this on my netwinder.  It also seems very strange that you 
> should be seeing this on a --disable-checking build but not if you apply 
> the patch and then enable checking.
> 
> Can you get a back-trace?
> 
> R.
> 

This is probably due to the fact that I'm (as always) optimizing for xscale ?
Oh, and I used '--without-checking' in stead of '--disable-checking', but that
should not make a difference.

I'm currently running the testsuite on my development board. (for gcc main, with patch).

When this is done, I'll revert the patch, and try again...
(hmm... this is a good way to keep a 733mhz xscale busy ;) )

Greetings,
-- 
Jeroen Dobbelaere
Embedded Software Engineer

ACUNIA Embedded Solutions
http://www.acunia.com/aes

Re: [patch] Re: Fix arm-linux bootstrap problem on cvs.

[Andreas Schwab]

jeroen dobbelaere <jeroen.dobbelaere@acunia.com> writes:

|> This is probably due to the fact that I'm (as always) optimizing for xscale ?
|> Oh, and I used '--without-checking' in stead of '--disable-checking', but that
|> should not make a difference.

--without-checking and --disable-checking are quite different things, and
the former is a no-op in gcc (it does not have a --with-checking configure
option).  To negate a --enable-foo configure option use --disable-foo or
--enable-foo=no (both being completely equivalent).

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux AG, Deutschherrnstr. 15-19, D-90429 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: [patch] Re: Fix arm-linux bootstrap problem on cvs.

[jeroen dobbelaere]

Andreas Schwab wrote:
> jeroen dobbelaere <jeroen.dobbelaere@acunia.com> writes:
> 
> |> This is probably due to the fact that I'm (as always) optimizing for xscale ?
> |> Oh, and I used '--without-checking' in stead of '--disable-checking', but that
> |> should not make a difference.
> 
> --without-checking and --disable-checking are quite different things, and
> the former is a no-op in gcc (it does not have a --with-checking configure
> option).  To negate a --enable-foo configure option use --disable-foo or
> --enable-foo=no (both being completely equivalent).
> 
> Andreas.
> 

oops
-- 
Jeroen Dobbelaere
Embedded Software Engineer

ACUNIA Embedded Solutions
http://www.acunia.com/aes

Re: [patch] Re: Fix arm-linux bootstrap problem on cvs.

[jeroen dobbelaere]

Richard Earnshaw wrote:
> jeroen.dobbelaere@acunia.com said:
> 
>>Bootstrapping the cvs without this patch (with --disable-checking)
>>failed :
> 
> 
> 
> Can't reproduce this on my netwinder.  It also seems very strange that you 
> should be seeing this on a --disable-checking build but not if you apply 
> the patch and then enable checking.
> 
> Can you get a back-trace?
> 
> R.
> 

This time, I did a build with :

cat config.status

#!/bin/sh
# This file was generated automatically by configure.  Do not edit.
# This directory was configured as follows:
../gcc/configure --with-gcc-version-trigger=/usr/src/cvs-others/gcc-trees/gcc-head/gcc/gcc/version.c --host=armv5l-unknown-linux-gnu --prefix=/usr/local/gcc-3.2-cvs --with-cpu=xscale --disable-checking --norecursion
#  using "mt-frag"


Again ending in :

/usr/src/cvs-others/gcc-trees/gcc-head/build/gcc/xgcc -shared-libgcc -B/usr/src/cvs-others/gcc-trees/gcc-head/build/gcc/ -nostdinc++ -L/usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/src 
-L/usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/src/.libs -B/usr/local/gcc-3.2-cvs/armv5l-unknown-linux-gnu/bin/ -B/usr/local/gcc-3.2-cvs/armv5l-unknown-linux-gnu/lib/ -isystem 
/usr/local/gcc-3.2-cvs/armv5l-unknown-linux-gnu/include -nostdinc++ -I/usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/include/armv5l-unknown-linux-gnu 
-I/usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/include -I../../../../gcc/libstdc++-v3/libsupc++ -I../../../../gcc/libstdc++-v3/libmath -g -O2 -D_GNU_SOURCE -fno-implicit-templates -Wall -Wno-format -W 
-Wwrite-strings -Winline -fdiagnostics-show-location=once -ffunction-sections -fdata-sections -g -c ../../../../gcc/libstdc++-v3/src/complex_io.cc  -fPIC -DPIC -o .libs/complex_io.o
In file included from /usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/include/ios:49,
                  from /usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/include/istream:44,
                  from /usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/include/sstream:44,
                  from /usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/include/complex:51,
                  from ../../../../gcc/libstdc++-v3/src/complex_io.cc:30:
/usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/include/bits/ios_base.h:427: internal compiler error: internal
    error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://www.gnu.org/software/gcc/bugs.html> for instructions.
make[4]: *** [complex_io.lo] Error 1
make[4]: Leaving directory `/usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/src'
make[3]: *** [all-recursive] Error 1
make[3]: Leaving directory `/usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3'
make[2]: *** [all-recursive-am] Error 2
make[2]: Leaving directory `/usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3'
make[1]: *** [all-target-libstdc++-v3] Error 2
make[1]: Leaving directory `/usr/src/cvs-others/gcc-trees/gcc-head/build'
make: *** [bootstrap] Error 2






Program received signal SIGSEGV, Segmentation fault.
0x001eef74 in ggc_set_mark (p=0x20) at ../../gcc/gcc/ggc-page.c:436
436 
   L2 = LOOKUP_L2 (p);
(gdb) where
#0  0x001eef74 in ggc_set_mark (p=0x20) at ../../gcc/gcc/ggc-page.c:436
#1  0x00029124 in gt_ggc_mx_lang_tree_node (x_p=0x20) at gtype-cp.h:265
(gdb) print p
$1 = (const void *) 0x20
(gdb) list
431 
   base = &table->table[0];
432 
#endif
433 

434 
   /* Extract the level 1 and 2 indices.  */
435 
   L1 = LOOKUP_L1 (p);
436 
   L2 = LOOKUP_L2 (p);
437 

438 
   return base[L1][L2];
439 
}
440 

(gdb)

Next thing I'll try is with the patch, but without the checking...


Greetings,
-- 
Jeroen Dobbelaere
Embedded Software Engineer

ACUNIA Embedded Solutions
http://www.acunia.com/aes

Re: [patch] Re: Fix arm-linux bootstrap problem on cvs.

[jeroen dobbelaere]

I get the same behavior with my patch. (No regressions or solutions there)

jeroen dobbelaere wrote:
[..]
> cat config.status
> 
> #!/bin/sh
> # This file was generated automatically by configure.  Do not edit.
> # This directory was configured as follows:
> ../gcc/configure 
> --with-gcc-version-trigger=/usr/src/cvs-others/gcc-trees/gcc-head/gcc/gcc/version.c 
> --host=armv5l-unknown-linux-gnu --prefix=/usr/local/gcc-3.2-cvs 
> --with-cpu=xscale --disable-checking --norecursion
> #  using "mt-frag"
> 
> 
> Again ending in :
> 
> /usr/src/cvs-others/gcc-trees/gcc-head/build/gcc/xgcc -shared-libgcc 
> -B/usr/src/cvs-others/gcc-trees/gcc-head/build/gcc/ -nostdinc++ 
> -L/usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/src 
> -L/usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/src/.libs 
> -B/usr/local/gcc-3.2-cvs/armv5l-unknown-linux-gnu/bin/ 
> -B/usr/local/gcc-3.2-cvs/armv5l-unknown-linux-gnu/lib/ -isystem 
> /usr/local/gcc-3.2-cvs/armv5l-unknown-linux-gnu/include -nostdinc++ 
> -I/usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/include/armv5l-unknown-linux-gnu 
> -I/usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/include 
> -I../../../../gcc/libstdc++-v3/libsupc++ 
> -I../../../../gcc/libstdc++-v3/libmath -g -O2 -D_GNU_SOURCE 
> -fno-implicit-templates -Wall -Wno-format -W -Wwrite-strings -Winline 
> -fdiagnostics-show-location=once -ffunction-sections -fdata-sections -g 
> -c ../../../../gcc/libstdc++-v3/src/complex_io.cc  -fPIC -DPIC -o 
> .libs/complex_io.o
> In file included from 
> /usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/include/ios:49, 
> 
>                  from 
> /usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/include/istream:44, 
> 
>                  from 
> /usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/include/sstream:44, 
> 
>                  from 
> /usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/include/complex:51, 
> 
>                  from ../../../../gcc/libstdc++-v3/src/complex_io.cc:30:
> /usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/include/bits/ios_base.h:427: 
> internal compiler error: internal
>    error: Segmentation fault
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See <URL:http://www.gnu.org/software/gcc/bugs.html> for instructions.
> make[4]: *** [complex_io.lo] Error 1
> make[4]: Leaving directory 
> `/usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3/src' 
> 
> make[3]: *** [all-recursive] Error 1
> make[3]: Leaving directory 
> `/usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3' 
> 
> make[2]: *** [all-recursive-am] Error 2
> make[2]: Leaving directory 
> `/usr/src/cvs-others/gcc-trees/gcc-head/build/armv5l-unknown-linux-gnu/libstdc++-v3' 
> 
> make[1]: *** [all-target-libstdc++-v3] Error 2
> make[1]: Leaving directory `/usr/src/cvs-others/gcc-trees/gcc-head/build'
> make: *** [bootstrap] Error 2
> 
> 
> 
> 
> 
> 
> Program received signal SIGSEGV, Segmentation fault.
> 0x001eef74 in ggc_set_mark (p=0x20) at ../../gcc/gcc/ggc-page.c:436
> 436   L2 = LOOKUP_L2 (p);
> (gdb) where
> #0  0x001eef74 in ggc_set_mark (p=0x20) at ../../gcc/gcc/ggc-page.c:436
> #1  0x00029124 in gt_ggc_mx_lang_tree_node (x_p=0x20) at gtype-cp.h:265
> (gdb) print p
> $1 = (const void *) 0x20
> (gdb) list
> 431   base = &table->table[0];
> 432 #endif
> 433
> 434   /* Extract the level 1 and 2 indices.  */
> 435   L1 = LOOKUP_L1 (p);
> 436   L2 = LOOKUP_L2 (p);
> 437
> 438   return base[L1][L2];
> 439 }
> 440
> (gdb)
> 
> Next thing I'll try is with the patch, but without the checking...
> 
> 
> Greetings,



-- 
Jeroen Dobbelaere
Embedded Software Engineer

ACUNIA Embedded Solutions
http://www.acunia.com/aes

Re: [patch] Re: Fix arm-linux bootstrap problem on cvs.

[Richard Earnshaw]

> I get the same behavior with my patch. (No regressions or solutions there)
> 
> jeroen dobbelaere wrote:

Ok, can you try a build with the following enabled:

--enable-checking=misc,tree,gc,rtl,rtlflag,gcac

It will make a rather slow compiler, but should enable us to determine 
whether this is a garbage collection problem.

R.

Re: [patch] Re: Fix arm-linux bootstrap problem on cvs.

[jeroen dobbelaere]

Richard Earnshaw wrote:
>>I get the same behavior with my patch. (No regressions or solutions there)
>>
>>jeroen dobbelaere wrote:
> 
> 
> Ok, can you try a build with the following enabled:
> 
> --enable-checking=misc,tree,gc,rtl,rtlflag,gcac
> 
> It will make a rather slow compiler, but should enable us to determine 
> whether this is a garbage collection problem.
> 
> R.
>

I've tried this, but I already did an gcc_update of the tree which seems to solve this
problem :

- I started one bootstrap without, the patch, with those checkings enabled.
   After it was somewhere in stage two (using the stage 1 compiler), I stopped
   it because it was trashing. (Only got 64M on my platform).

- I also did a bootstrap and testsuite for c,c++ on this CVS version with and
   without the patch.

   Results can be found here :
	<http://gcc.gnu.org/ml/gcc-testresults/2002-07/msg00323.html>

   and
	<http://gcc.gnu.org/ml/gcc-testresults/2002-07/msg00324.html>

Diff-ing those two results only shows differences in paths, not in testresults.

So I guess that this patch can be safely applied in order to enable cvs builds
with checking enabled.

Greetings
-- 
Jeroen Dobbelaere
Embedded Software Engineer

ACUNIA Embedded Solutions
http://www.acunia.com/aes

Re: [patch] Re: Fix arm-linux bootstrap problem on cvs.

[Richard Earnshaw]

> - I also did a bootstrap and testsuite for c,c++ on this CVS version with and
>    without the patch.
> 
>    Results can be found here :
> 	<http://gcc.gnu.org/ml/gcc-testresults/2002-07/msg00323.html>
> 
>    and
> 	<http://gcc.gnu.org/ml/gcc-testresults/2002-07/msg00324.html>
> 
> Diff-ing those two results only shows differences in paths, not in testresults.
> 
> So I guess that this patch can be safely applied in order to enable cvs builds
> with checking enabled.

Yep, thanks for your patience.   Please install.

R.

Re: [patch] Re: Fix arm-linux bootstrap problem on cvs.

[jeroen dobbelaere]

Richard Earnshaw wrote:
>>- I also did a bootstrap and testsuite for c,c++ on this CVS version with and
>>   without the patch.
>>
>>   Results can be found here :
>>	<http://gcc.gnu.org/ml/gcc-testresults/2002-07/msg00323.html>
>>
>>   and
>>	<http://gcc.gnu.org/ml/gcc-testresults/2002-07/msg00324.html>
>>
>>Diff-ing those two results only shows differences in paths, not in testresults.
>>
>>So I guess that this patch can be safely applied in order to enable cvs builds
>>with checking enabled.
> 
> 
> Yep, thanks for your patience.   Please install.
> 

I don't have write access to the tree.
Can someone who does install it ?

> R.
> 


Greetings,
-- 
Jeroen Dobbelaere
Embedded Software Engineer

ACUNIA Embedded Solutions
http://www.acunia.com/aes

Re: [patch] Re: Fix arm-linux bootstrap problem on cvs.

[Richard Earnshaw]

> Richard Earnshaw wrote:
> >>- I also did a bootstrap and testsuite for c,c++ on this CVS version with and
> >>   without the patch.
> >>
> >>   Results can be found here :
> >>	<http://gcc.gnu.org/ml/gcc-testresults/2002-07/msg00323.html>
> >>
> >>   and
> >>	<http://gcc.gnu.org/ml/gcc-testresults/2002-07/msg00324.html>
> >>
> >>Diff-ing those two results only shows differences in paths, not in testresults.
> >>
> >>So I guess that this patch can be safely applied in order to enable cvs builds
> >>with checking enabled.
> > 
> > 
> > Yep, thanks for your patience.   Please install.
> > 
> 
> I don't have write access to the tree.
> Can someone who does install it ?
> 

Ah!  Ok, I'll handle it.

Do you have a copyright assignment in place?  I don't think we need one 
for this change, after all each change is effectively to just one line and 
there are only two such changes, but it would be important to have one in 
place if you are to contribute regularly.

R.

Re: [patch] Re: Fix arm-linux bootstrap problem on cvs.

[jeroen dobbelaere]

Richard Earnshaw wrote:
[..]

>>
>>I don't have write access to the tree.
>>Can someone who does install it ?
>>
> 
> 
> Ah!  Ok, I'll handle it.
> 
> Do you have a copyright assignment in place?  I don't think we need one 

yep. That's ok.

> for this change, after all each change is effectively to just one line and 
> there are only two such changes, but it would be important to have one in 
> place if you are to contribute regularly.
> 
> R.
> 


Thanks and greetings,
-- 
Jeroen Dobbelaere
Embedded Software Engineer

ACUNIA Embedded Solutions
http://www.acunia.com/aes

Re: [patch] Re: Fix arm-linux bootstrap problem on cvs.

[Richard Earnshaw]

> > I don't have write access to the tree.
> > Can someone who does install it ?
> > 
> 
> Ah!  Ok, I'll handle it.

Applied.