Git Log

Commit Date  
[0d39b8] by yaberauneya yaberauneya

Commit execltp - a python script that provides a more user friendly alternative to runltp.

The only thing basic that's outstanding from this script that could be improved upon is making the while loop into a python iterator.

Signed-off-by: Garrett Cooper <>

2009-11-20 05:45:41 Tree
[651735] by yaberauneya yaberauneya

Append the PID to linux_syscall_numbers.h to avoid EBUSY / multiple file accesses with corrupt data at once if make(1) isn't doing the right thing and multiple jobs try to modify the header at the same time. This was based on personal experience with binutils and libtool modifying one file multiple times with parallel jobs via libtool.

Signed-off-by: Garrett Cooper <>

2009-11-19 19:35:15 Tree
[814189] by yaberauneya yaberauneya

Check in remaining syscall number modifications.

Signed-off-by: Garrett Cooper <>

2009-11-19 19:24:59 Tree
[db2e5c] by yaberauneya yaberauneya

This wasn't supposed to be checked in yet. Reverting...

Signed-off-by: Garrett Cooper <>

2009-11-19 19:10:04 Tree
[acb997] by yaberauneya yaberauneya

Include libevent in the proverbial mix.

Signed-off-by: Garrett Cooper <>

2009-11-19 19:08:36 Tree
[81db12] by yaberauneya yaberauneya

Fix more tail -<integer> syntax issues.

Signed-off-by: Garrett Cooper <>

2009-11-19 18:52:32 Tree
[50fb62] by yaberauneya yaberauneya

1. Merge and into one script - - where all you have to do is specify no arguments or 64 to execute ftruncate04 or ftruncate04_64, respectively.
2. Correct tail -1 syntax (-<int> argument types are deprecated with tail(1); Redhat barks whenever it executes it).

Signed-off-by: Garrett Cooper <>

2009-11-19 18:40:39 Tree
[e20782] by subrata_modak subrata_modak

mem/hugeshmget: fix failure of hugeshmget01: During hugeshmget01 testcase, we got the following:
hugeshmget01 1 TFAIL : seqment size is not correct
hugeshmget01 1 TFAIL : hugeshmget01 call failed - errno = 17 :
File exists ......
In shmget call, "buf.shm_segsz" had been set to "huge_pages_shm_to_be_allocated", but the code still used old macro "HUGE_SHM_SIZE", and this leaded to error "seqment size is not correct". So do the following change:
1) s/HUGE_SHM_SIZE/huge_pages_shm_to_be_allocated
2) Delete unused macro "HUGE_SHM_SIZE" defination
This patch fixed the failure. Signed-off-by: Liu Bo <>.

2009-11-19 16:05:21 Tree
[35b8cb] by yaberauneya yaberauneya

Hello guys,

I have run into some problems with the network stress tests. I submitted
a bug through the sourceforge bug tracker, but after subscribing to the
mailing list I can see that it's here where all the action is. If this
is not the right place, then please tell me what is the preferred way to
submit bug reports and patches.

I am using ltp-full-20091031.

A lot of the TCP tests involving packet loss fail with this error:

./tcp4-uni-basic01: line 394: netem_param: unbound variable

The reason is that strict checking for undefined variables is enabled
with "set -u", but the variable netem_param is not always defined before
testing its contents.

The fix I have applied is to initialize the netem_param variable with an
empty string.

Also if4-addr-change always fails with this error:

TBROK : Failed to set an IPv4 address at the remote host

I have found the reason to be a missing semicolon.

Best regards,
Radoslav Kolev

Signed-off-by: Garrett Cooper <>

2009-11-18 12:05:20 Tree
[0baf41] by subrata_modak subrata_modak

[PATCH (2/2)] splitting the tsc support. The patch moves the tsc macros form the librttest.h to another header file (named libtsc.h). Without this I got the following error and no tests were built.
In file included from librttest.c:43:
../include/librttest.h:127:2: error: #error
../include/librttest.h:169:2: error: #error
In file included from librttest.c:43:
../include/librttest.h: In function ‘atomic_add’:
The tsc macros header is only included in the following tests (thus where necessary):
o async_handler_tsc.c
o preempt_timing.c
o rdtsc-latency.c
Note1: instead of touching the make process the patch allows to build the tests above also for architecture that do not support tsc, yet. These tests will fail at run-time with ENOTSUP. A warning will appear while compiling as well. hmm, I do know if it is the right solution but it's simple and a good starting point for me. Signed-off-by: Giuseppe Cavallaro <>.

2009-11-16 13:30:05 Tree
[c3e457] by subrata_modak subrata_modak

[PATCH (1/2)] add atomi_add() define for __sh__. Signed-off-by: Giuseppe Cavallaro <>.

2009-11-16 13:28:58 Tree
[8eb5a8] by subrata_modak subrata_modak

mail: fix the bug of mail01: In the testcase "mail01", when mail is send to nosuchuser@domain and mail is send to user@nosuchdomain, mail program may return either "Returnedmail:" or "UndeliveredMailReturned", regardless of the existence of /etc/redhat-release. So, we did a bit of changes to recognize both of returned values. This patch fixes the problem. Signed-off-by: Liu Bo <>.

2009-11-16 13:25:48 Tree
[a6a802] by yaberauneya yaberauneya

Fix some remaining compile errors that we're checked in earlier...

Signed-off-by: Garrett Cooper <>

2009-11-16 08:41:06 Tree
[008a4c] by yaberauneya yaberauneya

1. Remove the .c files and all calls in because we can achieve the same thing through additional shell test logic.
2. Add the apicmds dependency.

Signed-off-by: Garrett Cooper <>

2009-11-16 08:36:37 Tree
[4c3ab3] by subrata_modak subrata_modak

mem/hugetlb: fix failure of hugemmap04: Testcase hugemmap04 needs at least one hugepage to test, so there needs a judgement of "number of hugepages". This patch fixed the problems. Signed-off-by: Liu Bo <>.

2009-11-16 08:01:15 Tree
[e2ac9f] by yaberauneya yaberauneya

As noticed by Shi Weihua, there was a syntax error in the awk command in get_cpu_usage:

awk: print $9
awk: ^ syntax error
pid 8077 cpu_usage
cpuctl_test_fj 1 TFAIL : case19 FAIL
awk: print $9
awk: ^ syntax error
pid 8104 cpu_usage
cpuctl_test_fj 1 TFAIL : case20 FAIL
awk: print $9
awk: ^ syntax error
pid 8131 cpu_usage
cpuctl_test_fj 1 TFAIL : case21 FAIL
awk: print $9
awk: ^ syntax error
pid 8193 cpu_usage
cpuctl_test_fj 1 TFAIL : case22 FAIL

The checkin corrects the syntax error by implementing the function in a much cleaner manner than the previous implementation, by using just ps | awk instead of top | tail | head | awk | awk.

Signed-off-by: Garrett Cooper <>

2009-11-16 07:40:40 Tree
[de3364] by yaberauneya yaberauneya

Let's not overwrite /bin/cat by accident in open08 if the setguid / setuid fails, mmk?

Signed-off-by: Garrett Cooper <>

2009-11-15 08:47:45 Tree
[da9635] by yaberauneya yaberauneya

1. Skip over /proc/irq by default to avoid broken hardware / drivers, and thus avoid `hangs'. You can traverse over /proc/irq using -i th
2. Make the selinux piece a compile-time conditional, to avoid the additional function call.
3. Fix the overflow issue on 64-bit systems by increasing values of total_obj, et all to unsigned long long's.
4. Only call close / closedir if fd is > 0 and dir is not NULL.

Signed-off-by: Garrett Cooper <>

2009-11-15 08:21:45 Tree
[ea695e] by yaberauneya yaberauneya

Checking in possible final / intermediate solution to ppoll01 not functioning on mips32. I say intermediate because of the following message:

AHA! I figured out why sigset is sized differently on mips32 for you,
compared to other architectures.

Just like there's a compat_caddr_t datatype (which is explicitly set
to char *), there's also a compat_sigset_t as well.

Let me see if I can figure out this mystery, but my guess is that
you're using n32 based mips, most likely on an Octeon model ;) ;)...
which would account for what you're seeing here and possibly account
for why the test consistently hangs on our Cavium boards at work (our
version of Linux has been hacked so bad though to take shortcuts in
the design process instead of following proper design procedure that
it makes me cry inside...).


The patch attached for ppoll01:

1. Cleans up spacing and code style
2. Removes unneeded/broken debug option parsing if test fails.
3. Prints out the last failed errno, rather than just the last errno(which may have been successful)
4. Passes in correct size of sigset_t for mips.

This patch was tested/passed on x86, mips(little endian), and ppc_82xx.

Signed-off-by: Henry Yei <>
Acked-by: Garrett Cooper <>

2009-11-15 01:17:43 Tree
[f9c2c5] by yaberauneya yaberauneya

I was REALLY stupid when I made clean dependent on ac-distclean. BAD BAD IDEA. Forcing people to have to call configure every time clean is run is STUPID design.

Signed-off-by: Garrett Cooper <>

2009-11-15 00:04:24 Tree
[3e8f77] by yaberauneya yaberauneya

1. Linker ordering is painful; -lltp must come last for some weird arse reason in order to work with binutils 1.19.51, whereas it needs to precede -lclone on 1.18.. We'll see whether or not it's fubar still in a minute...
2. Not all of the mq_* functions are syscalls; many are in fact libcalls, so we need to use the appropriate function call to stimulate them.

Signed-off-by: Garrett Cooper <>

2009-11-14 23:58:26 Tree
[60c8a6] by yaberauneya yaberauneya

Add the mq_* syscalls to the syscall list for each available architecture.

Signed-off-by: Garrett Cooper <>

2009-11-14 23:23:40 Tree
[be3fdc] by yaberauneya yaberauneya

Use linux_syscall_numbers.h instead of the syscalls. Need to add the __NR_ defs to the .in files still.

Signed-off-by: Garrett Cooper <>

2009-11-14 17:52:51 Tree
[4d462a] by yaberauneya yaberauneya

Change += back to := for LDLIBS. I did this for a reason (to make sure that everyone adds the appropriate LDLIBS values after including // //

We need to solve the linker issue with libclone, properly, not shove the issue under the rug with things like this.

Signed-off-by: Garrett Cooper <>

2009-11-14 10:29:25 Tree
[8680b2] by yaberauneya yaberauneya

Quell the _GNU_SOURCE already defined noise.

Signed-off-by: Garrett Cooper <>

2009-11-14 10:27:29 Tree
Older >