|
From: Simon J. B. <sim...@cm...> - 2004-06-02 02:39:50
|
I have tried configuring Valgrind for an AMD64 system with the following results checking for a supported version of gcc... ok (gcc (GCC) 3.3.2 (Mandrake Linux 10.0 3.3.2-6.1mdk)) checking build system type... x86_64-unknown-linux checking host system type... x86_64-unknown-linux checking for a supported CPU... no (x86_64) configure: error: Valgrind is ix86 specific. Sorry setting the host argument doesn't help. Is Valgrind compatible with non Intel architecture? Cheers, Simon |
|
From: Robert W. <rj...@du...> - 2004-06-02 02:47:05
|
> setting the host argument doesn't help. Is Valgrind compatible with non=20 > Intel architecture? Nope. Not yet, anyway. --=20 Robert Walsh Amalgamated Durables, Inc. - "We don't make the things you buy." Email: rj...@du... |
|
From: Andi K. <ak...@mu...> - 2004-06-03 17:13:25
|
"Simon J. Benson" <sim...@cm...> writes: > > setting the host argument doesn't help. Is Valgrind compatible with > non Intel architecture? Use linux32 ./configure ... However the result may not run completely reliable, there seem to be still some issues left in the 32bit emulation of the kernel interacting with valgrind Also of course you can only run 32bit executables this way. -Andi |
|
From: Robert W. <rj...@du...> - 2004-06-07 06:22:55
|
On Thu, 2004-06-03 at 10:13, Andi Kleen wrote: > "Simon J. Benson" <sim...@cm...> writes: > > > > setting the host argument doesn't help. Is Valgrind compatible with > > non Intel architecture? >=20 > Use linux32 ./configure ... >=20 > However the result may not run completely reliable, there seem > to be still some issues left in the 32bit emulation of the kernel > interacting with valgrind >=20 > Also of course you can only run 32bit executables this way. I've had some success running on AMD64 with the patch at this address: http://www.durables.org/software/valgrind/ This only works for 32-bit executables, and I built Valgrind on a P4 machine - I haven't tried the linux32 ./configure trick Andi mentioned. Regards, Robert. --=20 Robert Walsh Amalgamated Durables, Inc. - "We don't make the things you buy." Email: rj...@du... |
|
From: Andi K. <ak...@mu...> - 2004-06-07 11:50:27
|
Robert Walsh <rj...@du...> writes: > > I've had some success running on AMD64 with the patch at this address: > > http://www.durables.org/software/valgrind/ > > This only works for 32-bit executables, and I built Valgrind on a P4 > machine - I haven't tried the linux32 ./configure trick Andi mentioned. Ah that fixes the 4GB problem, right? I worked around that by using linux32 --3gb (which will the process only give 3GB address space) -Andi |
|
From: Bryan O'S. <bo...@se...> - 2004-06-07 18:03:25
|
On Mon, 2004-06-07 at 04:50, Andi Kleen wrote: > I worked around that by using linux32 --3gb (which will the process > only give 3GB address space) Trouble is, linux32 looks like it's only distributed by SuSE. I can't find anything similar in Red Hat land. <b |
|
From: Andi K. <ak...@mu...> - 2004-06-07 18:31:57
|
On Mon, Jun 07, 2004 at 11:03:16AM -0700, Bryan O'Sullivan wrote: > On Mon, 2004-06-07 at 04:50, Andi Kleen wrote: > > > I worked around that by using linux32 --3gb (which will the process > > only give 3GB address space) > > Trouble is, linux32 looks like it's only distributed by SuSE. I can't > find anything similar in Red Hat land. Debian should at least have it too. If you don't have it you can grab it from ftp://ftp.x86-64.org/pub/linux/tools/linux32/ Anyways - it would be better of course to fix valgrind to not assume 3GB address space. There's an open bug for this. -Andi |
|
From: Robert W. <rj...@du...> - 2004-06-07 18:36:01
|
> > I've had some success running on AMD64 with the patch at this address: > > > > http://www.durables.org/software/valgrind/ > > > > This only works for 32-bit executables, and I built Valgrind on a P4 > > machine - I haven't tried the linux32 ./configure trick Andi mentioned. > > Ah that fixes the 4GB problem, right? Um, yep. I don't remember the details right now, but Jeremy wasn't terribly happy with the fix as it potentially meant Valgrind could allocate some memory in the last page of the address space. He suggested another fix, but that caused a crash later in the code. I think you might find the thread in valgrind-developers somewhere. That's why I haven't checked this back in yet, BTW. Regards, Robert. |