I've been testing these against our entire kernel regression suite. While they definitely fix the original problems, there is one test that now hangs. I'm going to have to get to the bottom of that failure before I can sign off.
________________________________________________________________________ From: Paolo Bonzini <pbonz...@redhat.com<mailto:pbonz...@redhat.com>> Date: Tue Nov 11 2014 09:23:38 GMT-0500 (EST) To: qemu-devel@nongnu.org <qemu-devel@nongnu.org<mailto:qemu-devel@nongnu.org>> Cc: Richard Bilson <rbil...@qnx.com<mailto:rbil...@qnx.com>> Subject: [PATCH 0/3] apic: ExtINT fixes for QNX Richard Bilson reported some QNX emulation bugs. I sent him offlist a prototype patch, that he commented on. The first patch here works around what is arguably a QNX bug, that is triggered by the last two patches; but it is better to avoid spurious HLT wakeups, so here it is. The last two patches are what Richard had tested, amended with a fix for Windows XP + KVM + QEMU apic. Richard, can you confirm that these patches work? Paolo Paolo Bonzini (3): apic: avoid getting out of halted state on masked PIC interrupts apic: fix loss of IPI due to masked ExtINT apic: fix incorrect handling of ExtINT interrupts wrt processor priority hw/intc/apic.c | 20 ++++++++++++-------- target-i386/cpu.c | 8 ++++++-- 2 files changed, 18 insertions(+), 10 deletions(-) -- 2.1.0
