[Qemu-devel] Re: [PATCHv2 0/3] rwhandler: introduce and switch pci_host to it
On Mon, Jan 18, 2010 at 10:56 AM, Michael S. Tsirkin m...@redhat.com wrote: Alexander, so I assume the following patchset should be enough for you to implement u3 support, simply by creating your own rwhandler, and using pci_data_read/write directly there. I have pushed it to a temporary branch in my tree: git://git.kernel.org/pub/scm/linux/kernel/git/mst/qemu.git rwhandler Paul, any comments on this approach? I'll push this to my pci tree if this turns out to be helpful. Hope this helps, and sorry about the churn. I proposed earlier something similar for MMIO. The thread could be interesting: http://lists.gnu.org/archive/html/qemu-devel/2009-05/msg00095.html
[Qemu-devel] Re: [PATCHv2 0/3] rwhandler: introduce and switch pci_host to it
On Mon, Jan 18, 2010 at 05:50:35PM +, Blue Swirl wrote: On Mon, Jan 18, 2010 at 10:56 AM, Michael S. Tsirkin m...@redhat.com wrote: Alexander, so I assume the following patchset should be enough for you to implement u3 support, simply by creating your own rwhandler, and using pci_data_read/write directly there. I have pushed it to a temporary branch in my tree: git://git.kernel.org/pub/scm/linux/kernel/git/mst/qemu.git rwhandler Paul, any comments on this approach? I'll push this to my pci tree if this turns out to be helpful. Hope this helps, and sorry about the churn. I proposed earlier something similar for MMIO. The thread could be interesting: http://lists.gnu.org/archive/html/qemu-devel/2009-05/msg00095.html IIUC that patch seems to do much more, and also seems to involve all io? This one is just a library that devices can use. Intended users are not performance-critical like pci config. -- MST
[Qemu-devel] Re: [PATCHv2 0/3] rwhandler: introduce and switch pci_host to it
On Mon, Jan 18, 2010 at 07:53:25PM +0200, Michael S. Tsirkin wrote: On Mon, Jan 18, 2010 at 05:50:35PM +, Blue Swirl wrote: On Mon, Jan 18, 2010 at 10:56 AM, Michael S. Tsirkin m...@redhat.com wrote: Alexander, so I assume the following patchset should be enough for you to implement u3 support, simply by creating your own rwhandler, and using pci_data_read/write directly there. I have pushed it to a temporary branch in my tree: git://git.kernel.org/pub/scm/linux/kernel/git/mst/qemu.git rwhandler Paul, any comments on this approach? I'll push this to my pci tree if this turns out to be helpful. Hope this helps, and sorry about the churn. I proposed earlier something similar for MMIO. The thread could be interesting: http://lists.gnu.org/archive/html/qemu-devel/2009-05/msg00095.html IIUC that patch seems to do much more, and also seems to involve all io? This one is just a library that devices can use. Intended users are not performance-critical like pci config. To put it in other words: could you please be more explicit please? I failed to see the relevance ... -- MST
[Qemu-devel] Re: [PATCHv2 0/3] rwhandler: introduce and switch pci_host to it
On Mon, Jan 18, 2010 at 7:41 PM, Michael S. Tsirkin m...@redhat.com wrote: On Mon, Jan 18, 2010 at 07:53:25PM +0200, Michael S. Tsirkin wrote: On Mon, Jan 18, 2010 at 05:50:35PM +, Blue Swirl wrote: On Mon, Jan 18, 2010 at 10:56 AM, Michael S. Tsirkin m...@redhat.com wrote: Alexander, so I assume the following patchset should be enough for you to implement u3 support, simply by creating your own rwhandler, and using pci_data_read/write directly there. I have pushed it to a temporary branch in my tree: git://git.kernel.org/pub/scm/linux/kernel/git/mst/qemu.git rwhandler Paul, any comments on this approach? I'll push this to my pci tree if this turns out to be helpful. Hope this helps, and sorry about the churn. I proposed earlier something similar for MMIO. The thread could be interesting: http://lists.gnu.org/archive/html/qemu-devel/2009-05/msg00095.html IIUC that patch seems to do much more, and also seems to involve all io? This one is just a library that devices can use. Intended users are not performance-critical like pci config. To put it in other words: could you please be more explicit please? I failed to see the relevance ... Well, I thought the stuff I proposed earlier may be useful for comparison. It was also opt-in like yours, by the way. Also the end result that Paul wasn't interested in my patch may (or may not) be relevant with yours too.