On 15.03.2007 [12:28:50 -0700], Nishanth Aravamudan wrote: > > While running `make func` on ia64, I found that > truncate_sigbus_versus_oom was adding a negative value to a pointer and > causing a SEGV. This was because hpage_size was of type int, even though > the return type of gethugepagesize() is long. The multiplication was > truncated and led to negative values. Fix this throughout the tests, > just in case, by making hpage_size always a long.
So, as of this stack of 3 patches, on top of the current master branch, we have full ia64 support, as far as it can go right now: 1) malloc works (although the heap will always end up at 2^63 due to vhpt short mode). 2) functional tests that should run, PASS (although I'm seeing some odd behavior with private, where every so often it takes *forever* on the COW fault. Not exactly sure why it's inconsistent (I would expect it to be forever every time). 3) segment remapping is unsupported, due the same vhpt short mode. hugepages can only be used at 2^63 or higher, but obviously a programs segments won't necessarily be there. As David said, there might be ways around this, by leaving some data in small pages, but putting as much as possible in large pages. Thanks, Nish -- Nishanth Aravamudan <[EMAIL PROTECTED]> IBM Linux Technology Center ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Libhugetlbfs-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/libhugetlbfs-devel
