ia64 kernel doesn't validate memory attributes (WB, UC, etc) correctly for mmap of /dev/mem, /sys/.../legacy_mem, and reading /sys/.../rom. Please incorporate the patches here: https://enterprise.redhat.com/issue-tracker/?module=issues&action=view&tid=115881
please attach them directly to bugzilla. (Or better yet, get them upstream before .21?)
Created attachment 151665 [details] backport ioremap_page_range() Tony put these changes in his test tree,so they will go upstream shortly after 2.6.21. Apply these in this order: ioremap ia64-ioremap-align ia64-ioremap ia64-legacy_mem-wb ia64-mmap-validation
Created attachment 151666 [details] use same variable names as i386
Created attachment 151667 [details] use ioremap_page_range() to avoid unsupported attributes
Created attachment 151668 [details] allow WB mmap of legacy_mem (for option ROMs)
Created attachment 151669 [details] fail mmaps that span incompatible attributes
Doesn't apply to Linus' latest tree. + echo 'Patch #400 (linux-2.6-backport-ioremap-page-range.patch):' Patch #400 (linux-2.6-backport-ioremap-page-range.patch): + patch -p1 -s 1 out of 1 hunk FAILED -- saving rejects to file include/linux/io.h.rej The next patch would create the file lib/ioremap.c, which already exists! Assume -R? [n] Looks like these are 2.6.18 patches?
These are against RHEL5. What base would you like?
(In reply to comment #8) > These are against RHEL5. What base would you like? Seeing how this is a Fedora bug -- Fedora or upstream :) P.
Fedora constantly tracks upstream. Right now, it's at 2.6.21rc5-git12 for example.
OK. If Fedora tracks upstream closely, I don't think there's any rush. I'm happy to wait until Tony pushes them upstream and Fedora updates to an upstream release that includes them. I was under the mistaken impression that Fedora would be basically a RHEL5 kernel with some new patches that might be candidates for future RHEL5 updates. If you want to close this as WONTFIX or DEFERRED or whatever, that's fine with me. Sorry for the distraction.