On Wed, Feb 25, 2015 at 10:02 PM, Yinghai Lu <yinghai@xxxxxxxxxx> wrote: > On Wed, Feb 25, 2015 at 4:28 PM, Bjorn Helgaas <bhelgaas@xxxxxxxxxx> wrote: >> On Thu, Feb 19, 2015 at 03:40:41PM -0800, Paul Johnson wrote: >> >> Right now, all I know is "mpt2sas worked in 3.4, it stopped working in >> 3.4.1, and this patch fixes it." >> >> What I don't know is what changed in 3.4.1 to break it. > > We enable auto detection to enable realloc automatically for SRIOV case. > > and after pci realloc pci bar for lsi get new assigned values. > > assume even 3.4 if passing pci=realloc will make lsi card stop working. Can you be any more specific? What I'm looking for is: - this worked at commit X - commit Y broke it - commit Z fixes it That way we can connect the dots so people backporting patches know that if they backport Y, they should also backport Z along with it. And we can get warm fuzzies by double-checking that the 3.4 kernel Paul is running does not have Y but 3.4.1 does. Workaround info, e.g., "if you have commit Y but not Z, 'pci=XXX' is a workaround", would also be useful for the changelog. It would also be nice to know why Linux trips over this but not Windows, e.g., "Windows doesn't see this because it doesn't allocate more space for SR-IOV" or whatever. Bjorn -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
