From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.2b) Gecko/20021029 Phoenix/0.4 Description of problem: buf[] is never initialized; if for whatever reason, XFree86 is unable to open /proc/version, the read (which does initialize buf) will fail. While the "OS Kernel: " string does check to see if buf[] is NULL, buf[0] will only be 0 by chance. Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1. Mess with the kernel so that /proc/version doesn't exist/isn't readable 2. Run XFree86. 3. Notice the output of "OS Kernel:" is garbage. Actual Results: OS Kernel: (heart) PF Expected Results: OS Kernel: 2.4.18-14 Additional info:
I've created a fix for this which will be in rawhide CVS build 20021208.1 and later. Once it is confirmed to work correctly, I'll put it into the erratum branch also. Would you willing to test the rawhide build?
I'd certainly be willing to test it if I can find the time. It'd be easiest if I could take a patch to 4.2.0; if not, could you tip me towards where I need to get the version you'd like me to test? I've never retrieved anything from RawHide before.
Sure, rawhide is available from: ftp://rawhide.redhat.com/pub/redhat/linux/rawhide Also, I put up XFree86 snapshots, and beta test packages at: ftp://people.redhat.com/mharris/testing for different releases of Red Hat Linux. The extremely-unstable-blah-blah dir is the developmental stuff that ends up going into rawhide. (The directory name is named as such to discourage people from upgrading to these packages and expecting rock solid support). I haven't seen any problems yet being reported since I put this patch in, so I might generate one for 4.2.1 soon as well.
XFree86-4.2.1-10.3 has the backported fix. Please test and let me know if it works for you ok. ftp://people.redhat.com/mharris/testing/8.0/XFree86
Seems to work ok for me, closing as fixed in RAWHIDE.
Reclosing as fixed in ERRATA for upcoming erratum release.