On Thu, Apr 11, 2013 at 6:33 PM, Gavin Shan <shangw@xxxxxxxxxxxxxxxxxx> wrote: > On Fri, Apr 05, 2013 at 11:01:18AM +0800, Gavin Shan wrote: >>On Fri, Apr 05, 2013 at 10:54:29AM +0800, Gavin Shan wrote: >> >>The subject should be "[0/4]" instead of "[0/5]". Forgot to change >>that accordingly. Sorry for the confusion ;-) >> > > Bjorn, did you get change taking a look on this? Please let me know > if you have more comments :-) I did look at it; it looks good, and I made some more cleanups, which I just posted as a "v4" series. Take a look and see if it makes sense to you. Bjorn >>>While we setup MSI or MSI-X for specific PCI device, the address of >>>MSI or MSI-X capability structure is figured out from the config >>>space for multiple times. That's unnecessary and the patchset addresses >>>that. With the patchset applied, the latency for MSI or MSI-X setup >>>would be decreased hopefully. >>> >>>v2 -> v3: >>> * Remove checking on MSI/MSI-X capability offset in msi_set_enable() >>> and msix_set_enable(). Let the caller does the check >>> * Using "u8" for MSI/MSI-X/PM capability >>> * Let caller of pci_msi_check_device() checks MSI/MSI-X capability >>> * Replace msi_control_reg() with "pos + PCI_MSI_FLAGS" >>> * Merge [4/5] to [3/5] >>>v1 -> v2: >>> * Cache the MSI/MSI-X capability offset to pci_dev directly according >>> to Bjorn's suggestion. >>> * Rebase to 3.9.RC5 >>> >>>--- >>> >>>drivers/pci/msi.c | 110 +++++++++++++++++++++++---------------------------- >>>drivers/pci/msi.h | 1 - >>>include/linux/pci.h | 4 +- >>>3 files changed, 52 insertions(+), 63 deletions(-) >>> >>>Thanks, >>>Gavin >>> > -- 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
