On Wed, Jul 10, 2013 at 02:57:48PM +0200, Michal Privoznik wrote:
On 03.07.2013 17:29, Daniel P. Berrange wrote:
On Tue, Jul 02, 2013 at 05:53:05PM +0200, Michal Privoznik wrote:
The chardev alias assignment is going to be needed in a separate places, so it should be moved into a separate function rather than copying code randomly around. --- src/qemu/qemu_command.c | 75 +++++++++++++++++++++++++++++++++++++++++++------ src/qemu/qemu_command.h | 3 ++ 2 files changed, 70 insertions(+), 8 deletions(-)
diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c index ba93233..903839f 100644 --- a/src/qemu/qemu_command.c +++ b/src/qemu/qemu_command.c @@ -892,6 +892,65 @@ qemuAssignDeviceControllerAlias(virDomainControllerDefPtr controller) return 0; }
+int +qemuAssignDeviceChrAlias(virDomainDefPtr def, + virDomainChrDefPtr chr, + ssize_t idx) +{ + const char *prefix = NULL; + const char *prefix2 = NULL; + + switch ((enum virDomainChrDeviceType) chr->deviceType) { + case VIR_DOMAIN_CHR_DEVICE_TYPE_PARALLEL: + prefix = "parallel"; + break; + + case VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL: + prefix = "serial"; + break; + + case VIR_DOMAIN_CHR_DEVICE_TYPE_CONSOLE: + prefix = "console"; + prefix2 = "serial"; + break; + + case VIR_DOMAIN_CHR_DEVICE_TYPE_CHANNEL: + prefix = "channel"; + break; + + case VIR_DOMAIN_CHR_DEVICE_TYPE_LAST: + return -1; + } + + if (idx == -1) { + virDomainChrDefPtr **arrPtr; + size_t *cntPtr; + size_t i; + idx = 0; + + virDomainChrGetDomainPtrs(def, chr, &arrPtr, &cntPtr); + + for (i = 0; i < *cntPtr; i++) { + int thisidx; + if (((thisidx = qemuDomainDeviceAliasIndex(&(*arrPtr)[i]->info, prefix)) < 0) && + (prefix2 && + (thisidx = qemuDomainDeviceAliasIndex(&(*arrPtr)[i]->info, prefix2)) < 0)) { + virReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("Unable to determine device index for character device")); + return -1; + } + if (thisidx >= idx) + idx = thisidx + 1; + } + }
The commit message describes this as a simple refactoring, but this if (idx== -1) {...} is all new functionality compared to what is being replaced. I'm not too sure that this logic is correct either when dealing with <console> with a 'serialXX' alias.
This function just imitates other functions we've already: qemuAssignDeviceRedirdevAlias, qemuAssignDeviceHostdevAlias being examples. One can find even more.
Sure, that's not what I was complaining about though. The issue is that you're mixing plain refactoring of code, with the inclusion of extra functionality. Nothing in this patch ever passes a value 'idx == -1' so code to handle that scenario is not related to refactoring. It can be introduced in whatever patch actually needs that code.
And regarding <console> I've sent 2 patches, none of them was accepted. So I had to workaround the problem in my patch. What's the solution you're suggesting? Just to refresh our memory: for <console> the alias can be either 'consoleX' or 'serialX' depending if the daemon was restarted or not.
I described what I think needs fixing here: https://www.redhat.com/archives/libvir-list/2013-July/msg00107.html
IIUC, this patch is intended to change things so that after libvirtd is restarted, we get:
def->seriales[0]->info == "serial0" def->consoles[0]->info == "console0"
but this is fixing the wrong thing. There is only one physical device emulated in the guest, which is a serial port with id==serial0, and this is reflected correctly in the XML we generate. Only the internal struct is different.
So what needs fixing is the code which populated def->consoles[0]->info with "console0" instead of the correct "serial0" string at VM startup.
Daniel -- |: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :| |: http://libvirt.org -o- http://virt-manager.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: http://entangle-photo.org -o- http://live.gnome.org/gtk-vnc :|