Just quoting from our bug report:
As noted in the SUSE bug report, numerous /tmp race conditions exist in ppc64-diag, in particular:
rtas_errd/diag_support.c:233: char command[]="/usr/bin/find /proc/device-tree -name status -print > /tmp/get_dt_files";
rtas_errd/diag_support.c:241: fp1 = fopen("/tmp/get_dt_files", "r");
rtas_errd/prrn_hotplug:8:TMPFILE=`mktemp -p /tmp`
scripts/ppc64_diag_mkrsrc:126:mkdir "/tmp/diagSEsnap", 0775;
scripts/ppc64_diag_mkrsrc:127:$general_eed_file = "/tmp/diagSEsnap/snapH.tar.gz";
In the case of rtas_errd/prrn_hotplug, mktemp is used but is assumed to have succeeded; there is no check for the return value.
mktemp should probably be used properly in all of these. I don't know if the data in /tmp/diagSEsnap is sensitive or not, but if it is, the permissions on that directory should probably be tightened up.
I think a single CVE should suffice for this. The above is from ppc64-diag-2.6.1.
Thanks.
References:
https://bugzilla.novell.com/show_bug.cgi?id=882667
https://bugzilla.redhat.com/show_bug.cgi?id=1109371
– Vincent Danen / Red Hat Product Security