Red Hat Bugzilla – Bug 16485
Problem with latest upgrade to mailx: core dump
Last modified: 2008-05-01 11:37:57 EDT
I have RH 6.2 installed. Using the upgrade agent, I upgraded my mailx-8.1.1-10.i386.rpm to mailx-8.1.1-16.i386.rpm. After the upgrade, whenever I run mail, open a message to view, and then exit mail, I get a segmentation fault and core dump. See the support incident #154690. I have uninstalled mailx-8.1.1-16.i386.rpm, and restored mailx-8.1.1-10.i386.rpm. Mail now works properly. I have core files I can email to you if necessary.
I have observed the same problem. It does not always core dump, but can be reliably made to do so by almost any activity in mail and then the 'q' command.
I got some free time and decided to look at this myself. The bug was introduced with the patch mailx-8.1.1-environ.patch. After applying this patch, the file fio.c contains case '&': if (name[1] == 0 && (name = value("MBOX")) == NOSTR) if (name[1] == 0 && (name = getenv("MBOX")) == NOSTR) name = "~/mbox"; /* fall through */ The first "if" results in name == NULL and the second if evaluates name[1]. Hence the core dump. It should look like case '&': if (name[1] == 0 && (name = value("MBOX")) == NOSTR) if ( (name = getenv("MBOX")) == NOSTR) name = "~/mbox"; /* fall through */ A diff for the patch is appended. Though patching patches may not be a good idea. John Franks john@math.nwu.edu --- mailx-8.1.1-environ.patch.orig Thu Aug 24 15:54:19 2000 +++ mailx-8.1.1-environ.patch Thu Aug 24 15:58:33 2000 @@ -61,7 +61,7 @@ return savestr(prevfile); case '&': if (name[1] == 0 && (name = value("MBOX")) == NOSTR) -+ if (name[1] == 0 && (name = getenv("MBOX")) == NOSTR) ++ if ((name = getenv("MBOX")) == NOSTR) name = "~/mbox"; /* fall through */ }
This is the same problem as bug #15728, fixed in Raw Hide but not in the Red Hat 6.2 errata.
Well, many of us are not permitted to see bug #15728. I should have mentioned that a workaround is to add the line set MBOX=~/mbox to your .mailrc file.
I have the same problem upgrading from mailx-8.1.1-10.i386.rpm and mailx-8.1.1-11.i386.rpm. I have many users do not want to remember yet another admin step. Can't mailx be officially re-released with the fix some time soon?
*** Bug 17535 has been marked as a duplicate of this bug. ***
*** Bug 17461 has been marked as a duplicate of this bug. ***
Fixes for bugs #14010, #15278, and #10074 were not included in the released errata. Re-releasing mailx-8.1.1-20
job@redhat.com wrote: > Re-releasing mailx-8.1.1-20 But it has not yet shown up a sa Red Hat 6.2 erratum. Wil it ever do so? Thanks, Jonathan Marsden <jonathan@xc.org>
If this is fixed then why is the lates version of mailx in the redhat updates directory still mailx-8.1.1-16.i386.rpm and mailx-8.1.1-16.src.rpm ?
Still no mailx-8.1.1-20 on the updates site in the SRPM or the i386 directories. Why has this been resolved?
I can find mailx-8.1.1-21.*.rpm from Polished linux on rpmfind network. Up2date wants to install mailx-8.1.1-16 when using mailx-8.1.1-10 from RH 6.2 unless the Polished version is installed. Once mailx-8.1.1-16 is installed, it just crashes.