Bug 249996

Summary:	linux/arm: unhandled syscall: 181 (__NR_pwrite64)
Product:	[Developer tools] valgrind	Reporter:	Peter Maydell <peter.maydell>
Component:	general	Assignee:	Julian Seward <jseward>
Status:	RESOLVED FIXED
Severity:	normal	CC:	Jacob.Bramley+kde
Priority:	NOR
Version:	3.6 SVN
Target Milestone:	---
Platform:	Compiled Sources
OS:	Linux
Latest Commit:		Version Fixed In:
Sentry Crash Report:
Attachments:	full valgrind -v log

Description Peter Maydell 2010-09-03 13:03:42 UTC

Created attachment 51265 [details]
full valgrind -v log

I've just tried running valgrind on firefox on this A8 pegatron. Mostly it works fine. However when I point the browser at maps.google.com valgrind complains:
./vg-in-place  -v --trace-children=yes firefox http://maps.google.com
[...]
--24646-- WARNING: unhandled syscall: 181
==24646==    at 0x4851FC7: __libc_do_syscall (libc-do-syscall.S:41)
==24646==    by 0x4850A94: pwrite64 (pwrite64.c:55)
==24646==    by 0x716FF4E: ??? (in /usr/lib/libtdb.so.1.2.1)
--24646-- You may be able to write your own handler.
--24646-- Read the file README_MISSING_SYSCALL_OR_IOCTL.
--24646-- Nevertheless we consider this a bug.  Please report
--24646-- it at http://valgrind.org/support/bug_reports.html.

unistd.h says 181 is __NR_pwrite64.

I'll attach the entirety of the log.

Comment 1 Julian Seward 2010-09-03 15:30:27 UTC

Does this fix it?  I can't test because I can't reproduce the
failure (glibc differences, possibly).

Index: coregrind/m_syswrap/syswrap-arm-linux.c
===================================================================
--- coregrind/m_syswrap/syswrap-arm-linux.c     (revision 11321)
+++ coregrind/m_syswrap/syswrap-arm-linux.c     (working copy)
@@ -1444,7 +1444,7 @@
    LINX_(__NR_rt_sigsuspend,     sys_rt_sigsuspend),  // 179
 
    GENXY(__NR_pread64,           sys_pread64),        // 180
-   //GENX_(__NR_pwrite64,          sys_pwrite64_on32bitplat),       // 181
+   GENX_(__NR_pwrite64,          sys_pwrite64),       // 181
    LINX_(__NR_chown,             sys_chown16),        // 182
    GENXY(__NR_getcwd,            sys_getcwd),         // 183
    LINXY(__NR_capget,            sys_capget),         // 184

Comment 2 Peter Maydell 2010-09-03 16:17:08 UTC

Yes, that change seems to fix things. It also causes openoffice to start properly (it was previously also complaining about this syscall being missing.) Thanks.

Comment 3 Julian Seward 2010-09-03 16:25:34 UTC

Change in comment #1 committed as r11334.

Comment 4 Tom Hughes 2011-08-10 12:46:21 UTC

*** Bug 225367 has been marked as a duplicate of this bug. ***