在 2019/8/20 上午4:51, Alex Williamson 写道:
On Thu, 15 Aug 2019 21:02:58 +0800
Ben Luo <[email protected]> wrote:
Currently, VFIO takes a lot of free-then-request-irq actions whenever
a VM (with device passthru via VFIO) sets irq affinity or mask/unmask
irq. Those actions only change the cookie data of irqaction or even
change nothing. The free-then-request-irq not only adds more latency,
but also increases the risk of losing interrupt, which may lead to a
VM hung forever in waiting for IO completion
What guest environment is generating this? Typically I don't see that
Windows or Linux guests bounce the interrupt configuration much.
Thanks,
Alex
By tracing centos5u8 on host, I found it keep masking and unmasking
interrupt like this:
[1566032533709879] index:28 irte_hi:000000010004a601
irte_lo:adb54bc000b98001
[1566032533711242] index:28 irte_hi:0000000000000000
irte_lo:0000000000000000
[1566032533711258] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533711269] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533711291] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533711321] index:28 irte_hi:0000000000000000
irte_lo:0000000000000000
[1566032533711340] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533711361] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533711376] index:28 irte_hi:000000010004a601
irte_lo:adb54bc000b98001
[1566032533713368] index:28 irte_hi:0000000000000000
irte_lo:0000000000000000
[1566032533713385] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533713396] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533713416] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533713447] index:28 irte_hi:0000000000000000
irte_lo:0000000000000000
[1566032533713464] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533713485] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533713499] index:28 irte_hi:000000010004a601
irte_lo:adb54bc000b98001
[1566032533718855] index:28 irte_hi:0000000000000000
irte_lo:0000000000000000
[1566032533718871] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533718882] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533718902] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533718932] index:28 irte_hi:0000000000000000
irte_lo:0000000000000000
[1566032533718949] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533718969] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533718984] index:28 irte_hi:000000010004a601
irte_lo:adb54bc000b98001
[1566032533719873] index:28 irte_hi:0000000000000000
irte_lo:0000000000000000
[1566032533719889] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533719900] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533719921] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533719954] index:28 irte_hi:0000000000000000
irte_lo:0000000000000000
[1566032533719971] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533719992] index:28 irte_hi:000000000004a601
irte_lo:00003fff00ac002d
[1566032533720007] index:28 irte_hi:000000010004a601
irte_lo:adb54bc000b98001
"[1566032533720007]" is timestamp in μs, so centos5u8 tiggers 30+ irte
modification within 10ms
Thanks,
Ben
This patchset solved the issue by:
Patch 2 introduces update_irq_devid to only update dev_id of irqaction
Patch 3 make use of update_irq_devid and optimize irq operations in VFIO
changes from v2:
- reformat to avoid quoted string split across lines and etc.
changes from v1:
- add Patch 1 to enhance error recovery etc. in free irq per tglx's comments
- enhance error recovery code and debugging info in update_irq_devid
- use __must_check in external referencing of update_irq_devid
- use EXPORT_SYMBOL_GPL for update_irq_devid
- reformat code of patch 3 for better readability
Ben Luo (3):
genirq: enhance error recovery code in free irq
genirq: introduce update_irq_devid()
vfio_pci: make use of update_irq_devid and optimize irq ops
drivers/vfio/pci/vfio_pci_intrs.c | 101 +++++++++++++++++++++-------------
include/linux/interrupt.h | 3 ++
kernel/irq/manage.c | 110 +++++++++++++++++++++++++++++++++-----
3 files changed, 164 insertions(+), 50 deletions(-)
