[Ltrace-devel] [PATCH] SH4 support

[Giuseppe Cavallaro]

I was able to run "make check" command on the SH4 target (but not using
dejagnu remotely) with the following results:
  - 74 tests passed
  - 6 tests failed
On my host I can see:
 - 79 passed
 - 1 failed

I've noticed ltrace for SH4 is not able to attach and trace (see the log
below) a process by pid (so the "attach-process" test fails).
Any idea? Where can I start looking at?

....
DEBUG: breakpoints.c:31: insert_breakpoint(): symbol=getpwuid, addr=0x4012a0
DEBUG: breakpoint.c:25: enable_breakpoint():
enable_breakpoint(2894,0x4012a0)
DEBUG: process_event.c:93: process_event(): event: signal (SIGSTOP [19])
2894 --- SIGSTOP (Stopped (signal)) ---
DEBUG: process_event.c:93: process_event(): event: signal (SIGSTOP [19])
2894 --- SIGSTOP (Stopped (signal)) ---
DEBUG: process_event.c:109: process_event(): event: syscall (SYS_write [4])
2894 SYS_write(1, "7 28-11.test\n", 13 <no return ...>
DEBUG: process_event.c:115: process_event(): event: sysret (SYS_write [4])
2894 <... SYS_write resumed> )                       = 13
DEBUG: process_event.c:119: process_event(): event: breakpoint
2894 sprintf( <unfinished ...>
DEBUG: breakpoints.c:31: insert_breakpoint(): symbol=(nil), addr=0x4041d8
DEBUG: breakpoint.c:25: enable_breakpoint():
enable_breakpoint(2894,0x4041d8)
DEBUG: process_event.c:119: process_event(): event: breakpoint
DEBUG: breakpoint.c:25: enable_breakpoint():
enable_breakpoint(2894,0x4016e4)
DEBUG: process_event.c:119: process_event(): event: breakpoint
2894 <... sprintf resumed>
"\2145W),\231\315{(CX)~U\335q\030\376\377\377@\337V)\370\231
\315{", "\2145W),\231\315{(CX)~U\335q\030"...) = 14
DEBUG: process_event.c:109: process_event(): event: syscall (SYS_exit [1])
2894 SYS_exit(2077072044 <no return ...>
DEBUG: process_event.c:115: process_event(): event: sysret (SYS_exit [1])
2894 <... SYS_exit resumed> )                        = <void>
DEBUG: process_event.c:109: process_event(): event: syscall (SYS_exit [1])
2894 SYS_exit(2077072044 <no return ...>
DEBUG: process_event.c:115: process_event(): event: sysret (SYS_exit [1])
2894 <... SYS_exit resumed> )                        = <void>
DEBUG: process_event.c:109: process_event(): event: syscall (SYS_exit [1])
2894 SYS_exit(2077072044 <no return ...>
...

Giuseppe

On 18/06/07, Giuseppe Cavallaro <peppe.cavallaro at gmail.com> wrote:
>
> On 14/06/07, Ian Wienand < ianw at gelato.unsw.edu.au> wrote:
> >
> > On Wed, Jun 13, 2007 at 03:59:49PM +0200, Giuseppe Cavallaro wrote:
> > > where can I download the right testsuite version for ltrace 0.4?
> > > So I'll provide the testsuite results.
> >
> > Please make sure your patch applies against SVN (it should be fine)
>
>
> in attachment  the new PATCH based on the latest tree.
>
> and then run the test suite with "make check"
>
>
> Concerning this point, it should take a long while.
> Unfortunately, I've not built and installed dejagnu package on my SH4
> target rootFS yet.
> So I cannot run "make check" command on the target, directly.
> I'd like to execute the testsuite using dejagnu working "cross-target
> mode".
> I've modified both the testsuite/Makefile and my dejagnu configuration
> files (site.exp etc...) so all tests are remotely built on SH4 target but
> the ltrace command, built for SH4  and invoked by dejagnu, is executed on
> the i386 host, so all tests fails (for more details see the log in
> attachment).
> How can I force dejagnu to execute the ltrace command on the remote
> target?
>
> Regards,
> Giuseppe
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.alioth.debian.org/pipermail/ltrace-devel/attachments/20070619/33a81c30/attachment.htm