On Sat, Dec 10, 2005 at 05:11:09PM +0100, Jakub Wilk wrote:
> >>>Could you please confirm that Debian bug #340659: "sed:
> >>>s/(.{30000})/\1/ -- segmentation fault" is fixed, at least in sed
> >>>4.1.4-4 from unstable?
> >>No, sed 4.1.4-4 still segfaults.
> >Thats strage; it works here. Could run it under valgrind or gdb?
> Sure, here's the valgrind's output (sed 4.1.4-4):
>
> #v+
> Conditional jump or move depends on uninitialised value(s)
> at 0x1B8ECB13: (within /lib/ld-2.3.5.so)
These are "expected" with libc235.
> Invalid read of size 4
> at 0x1B9B437B: (within /lib/tls/libc-2.3.5.so)
> by 0x1B9BF87C: (within /lib/tls/libc-2.3.5.so)
> by 0x1B9BFCFE: re_compile_pattern (in /lib/tls/libc-2.3.5.so)
> by 0x804E2C5: (within /bin/sed)
> by 0x804E3B5: (within /bin/sed)
> by 0x804B539: (within /bin/sed)
> by 0x804BB76: (within /bin/sed)
> by 0x804977D: (within /bin/sed)
> by 0x1B926ECF: __libc_start_main (in /lib/tls/libc-2.3.5.so)
> by 0x80493A0: (within /bin/sed)
> Address 0x1BFF5F60 is not stack'd, malloc'd or (recently) free'd
This is at least marginally useful. Would you consider compiling sed
locally with export DEB_BUILD_OPTIONS=nostrip set? That will allow a
better backtrace.
Also, Clint just uploaded 4.1.4-5, which closes this bug. I don't
know what he meant, "This may have been closed incorrectly."; I might
have caused some confusion by digging into the bugs while he was
preparing an upload. Anyway, could you test the new version? It is
not yet "installed" into the archive, so you'll have to use
<http://incoming.debian.org/>. Actually, 386 packages aren't even
available yet. You could wait a short while until they show up, or
build yourself from the .diff at incoming.d.o and the .orig at
ftp.d.o, or I could build it for you.
--
Clear skies,
Justin
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
