Messages in this thread | | | Date | Sun, 25 May 2008 16:37:57 -0600 | From | Grant Grundler <> | Subject | Re: [PATCH] x86: early dump pci conf space v2 |
| |
On Thu, May 22, 2008 at 11:38:24PM -0700, Yinghai Lu wrote: ... > > Having debugged lots of PCI resource and config issues, I understand > > why this is useful and do NOT object to it being included in the tree. > > could print out the setting from buggy BIOS, or first kernel. > and don't need the reporter apply debug patch again...
Yes, I understood. Here are some more details on why my first reaction was negative: o Most of the time the system can boot anyway and we get lspci -v output, /proc/iomem, etc. If that's not true, then other parts of the kernel need to be more robust (e.g. drivers crashing because resources aren't assigned.
o Bringing up linux on a new platform is NOT something we've generally had to do via email. Usually someone with kernel/IO experience is involved and can build new kernels.
> > At least if you want people to use it. :) > > Perhaps Jesse can suggest a preferred location. > > we could add it. > .. > > > > Please call this "pci_early_dump" (I prefer with underscores but don't care). > > Don't want it confused with kernel dumps (LKCD) or any other > > general debugging tools. > > so you want pci_early_dump instead of pci=earlydump? > or pci=early_dump?
Sorry, I misread the patch. I was thinking of "pci_early_dump" as a seperate option but either "pci=" option is fine.
thanks, grant
| |