[m5-dev] Cron m5t...@zizzer /z/m5/regression/do-regression quick

2009-08-07 Thread Cron Daemon
* build/ARM_SE/tests/fast/quick/00.hello/arm/linux/simple-atomic passed. * build/SPARC_SE/tests/fast/quick/40.m5threads-test-atomic/sparc/linux/simple-timing-mp passed. * build/SPARC_SE/tests/fast/quick/00.hello/sparc/linux/simple-timing passed. *

[m5-dev] using gdb with memtest-ruby

2009-08-07 Thread Somayeh Sardashti
Hi, I really need to use a debugging tool like gdb, but it seems that it just work with m5.debug. scons run memtest-ruby with m5.fast, how can I make it use m5.debug? Thanks, Somayeh ___ m5-dev mailing list m5-dev@m5sim.org

Re: [m5-dev] Bug in DATA_MESSAGE_SIZE initialization in Ruby

2009-08-07 Thread Krishna, Tushar
Hi Derek, Here is the patch of my changes. If it looks fine, I will run the regression test again with update_ref=True, send out the final patch, and push in the changes. Thanks Tushar

[m5-dev] gzstream library

2009-08-07 Thread Derek Hower
When you compile m5 as a static library, e.g. scons build/ALPHA_SE/libm5_prof.a it doesn't compile gzstream into the library, nor does it link against a separate gzstream library, making the m5 library unusable. How can I make it compile gzstream? -Derek

[m5-dev] changeset in m5: X86: Don't truncate the immediate parameter for...

2009-08-07 Thread Gabe Black
changeset adbd07f1630d in /z/repo/m5 details: http://repo.m5sim.org/m5?cmd=changeset;node=adbd07f1630d description: X86: Don't truncate the immediate parameter for the ENTER instruction. diffstat: 1 file changed, 2 insertions(+), 2 deletions(-)

[m5-dev] changeset in m5: X86: Use the right field when using legacy pref...

2009-08-07 Thread Gabe Black
changeset f3c9335ec2cd in /z/repo/m5 details: http://repo.m5sim.org/m5?cmd=changeset;node=f3c9335ec2cd description: X86: Use the right field when using legacy prefixes to distinguish instructions. diffstat: 1 file changed, 1 insertion(+), 1 deletion(-)

[m5-dev] changeset in m5: X86: Make the qaud width bswap instruction hand...

2009-08-07 Thread Gabe Black
changeset 2ec6bfc8f9c7 in /z/repo/m5 details: http://repo.m5sim.org/m5?cmd=changeset;node=2ec6bfc8f9c7 description: X86: Make the qaud width bswap instruction handle the fact that 32 bit operations zero extend. diffstat: 1 file changed, 10 insertions(+), 10 deletions(-)

[m5-dev] changeset in m5: X86: Implement shift right/left double microops.

2009-08-07 Thread Gabe Black
changeset b9ab1b56391b in /z/repo/m5 details: http://repo.m5sim.org/m5?cmd=changeset;node=b9ab1b56391b description: X86: Implement shift right/left double microops. This is my best guess as far as what these should do. Other existing microops use implicit registers, mul1s

[m5-dev] changeset in m5: X86: (Re)Implemented SHRD.

2009-08-07 Thread Gabe Black
changeset fa6d324aa2f9 in /z/repo/m5 details: http://repo.m5sim.org/m5?cmd=changeset;node=fa6d324aa2f9 description: X86: (Re)Implemented SHRD. diffstat: 2 files changed, 30 insertions(+), 19 deletions(-) src/arch/x86/isa/decoder/two_byte_opcodes.isa|2

Re: [m5-dev] gzstream library

2009-08-07 Thread nathan binkert
When you compile m5 as a static library, e.g. scons build/ALPHA_SE/libm5_prof.a it doesn't compile gzstream into the library, nor does it link against a separate gzstream library, making the m5 library unusable. How can I make it compile gzstream? There should be a libgzstream.a in

Re: [m5-dev] using gdb with memtest-ruby

2009-08-07 Thread nathan binkert
I really need to use a debugging tool like gdb, but it seems that it just work with m5.debug. scons run memtest-ruby with m5.fast, how can I make it use m5.debug? In your scons target, you must have the word fast in there somewhere. If you replace it with debug, it should work. If this

Re: [m5-dev] gzstream library

2009-08-07 Thread Derek Hower
On Fri, Aug 7, 2009 at 12:46 PM, nathan binkertn...@binkert.org wrote: When you compile m5 as a static library, e.g. scons build/ALPHA_SE/libm5_prof.a it doesn't compile gzstream into the library, nor does it link against a separate gzstream library, making the m5 library unusable. How can

Re: [m5-dev] gzstream library

2009-08-07 Thread Derek Hower
Command: g++ -o bochs -I/u/d/r/drh5/mfacet/rocks -I/u/d/r/drh5/mfacet/rocks/bochs -I/u/d/r/drh5/mfacet/rocks/gem5/src/mem/ruby -DBX_WITH_RUBY=1 -DBX_241=1 -g -pg -O2 -D_FILE_OFFSET_BITS=64 -D_LARGE_FILES -I/usr/include/SDL -D_GNU_SOURCE=1 -D_REENTRANT -DBX_SHARE_PATH=\/usr/local/share/bochs\

Re: [m5-dev] gzstream library

2009-08-07 Thread nathan binkert
I think this is because of the ordering on the command line. If I recall correctly, when two libraries depend on each other, the providing library needs to be listed *after* the consuming library. So you should put gzstream after m5. This is because with static libs linked with -l, the linker

Re: [m5-dev] using gdb with memtest-ruby

2009-08-07 Thread somayeh
Hi, When I run the debug version, it returns this error: m5.debug: build/ALPHA_SE/mem/ruby/eventqueue/RubyEventQueue.cc:79: void RubyEventQueue::scheduleEventAbsolute(Consumer*, Time): Assertion `timeAbs m_globalTime' failed. Program aborted at cycle 256 I did not have this problem when I was

Re: [m5-dev] using gdb with memtest-ruby

2009-08-07 Thread Steve Reinhardt
The reason you don't get the error with m5.fast is because m5.fast compiles out assertions. The issue is still there. (In general I like to run regressions with m5.opt rather than m5.fast for this reason.) Using gdb is easy if you run m5 directly and not via the scons regression framework

Re: [m5-dev] gzstream library

2009-08-07 Thread Derek Hower
Well that is handy to know. Seems to do the trick. Thanks, Nate So, now how can I get scons to build the gzstream library when you specify a static library target? On Fri, Aug 7, 2009 at 3:44 PM, nathan binkertn...@binkert.org wrote: I think this is because of the ordering on the command

Re: [m5-dev] gzstream library

2009-08-07 Thread nathan binkert
Well that is handy to know.  Seems to do the trick.  Thanks, Nate So, now how can I get scons to build the gzstream library when you specify a static library target? Well, you can just specify gzstream as a target: build/gzstream/libgzstream.a I think that the longer term solution is to just

Re: [m5-dev] [PATCH] bug fix for data_msg_size in network/Network.cc

2009-08-07 Thread nathan binkert
Unless anybody has any concerns, I will try and check in the changes to the repository in an hour. Tushar, I don't have a problem with your patch, but an hour is not giving people enough time to review your patch. If your code is in a mercurial queue, there should be no big rush and you