Bug 186520
Summary: | CONFIG_PCI_MSI is not set to y in installed x86_64 2.6.15 kernel for fc5 | ||
---|---|---|---|
Product: | [Fedora] Fedora | Reporter: | Dave Olson <dave.olson> |
Component: | kernel | Assignee: | Dave Jones <davej> |
Status: | CLOSED ERRATA | QA Contact: | Brian Brock <bbrock> |
Severity: | medium | Docs Contact: | |
Priority: | medium | ||
Version: | 5 | CC: | bos, lindahl, pfrields, rjwalsh, wtogami |
Target Milestone: | --- | ||
Target Release: | --- | ||
Hardware: | x86_64 | ||
OS: | Linux | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | Bug Fix | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2006-07-16 01:44:34 UTC | Type: | --- |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: |
Description
Dave Olson
2006-03-24 00:50:50 UTC
It caused problems booting on some systems. I don't have the details to hand. This is a big problem for us. 1. Our cards are the first of probably many which require MSI. MSI was optional for PCI-X but is required for PCI Express. 2. Other cards like the Mellanox InfiniBand adaptor run faster with MSI enabled -- see the OpenIB mailing list for discussion about this a while ago. I have no doubt that some systems have problems with MSI, but they're broken and will probably get fixed, for example I spent 2 days arguing with the in-house BIOS team at a Tier 1 computer vendor (a big Red Hat partner) and they finally agreed that they needed to fix their BIOS so that MSI worked. It turned out to be a one-line fix... It appears that the latest FC 4 2.6.16 kernel now has the same issue, that is, the default 2.6.16 kernel with FC4 yum update now has CONFIG_PCI_MSI=n It appears that the latest fc4 update kernels (e.g. 2.6.16-1.2108_FC4smp for x86_64) now has CONFIG_PSI_MSI=y again, and I'm told FC5 has also reverted to enabling this. Should we expect that to remain true for future Fedora kernels? no (it was on as an experiment in fc4, and sneaked into an update inadvertantly), I'm going to look at enabling it across the board when I rebase to 2.6.17 |