[desc]:
can use cmdline in grub to:
1). custom set sleep time after device reset.
e.g. pcireset_quirk=timeout,vid1:did1=t1,vid2:did2=t2
this will set t1 second sleep after vid1:did1 device reset and so do vid2:did2
use to solve zixiao gpu reset problem.
2). return 0 directly when device reset.
e.g. pcireset_quirk=noreset,vid1:did1,vid2:did2
this will return 0 when vid1:did1 device reset and so do vid2:did2
use to solve 2080Ti & 3070 GPU reset cause ff problem.
Signed-off-by: leolingli <leolingli@tencent.com>
Signed-off-by: snailzhao <snailzhao@tencent.com>
upstream: 51c48b310183ab6ba5419edfc6a8de889cc04521
pci_bridge_check_ranges() determines whether a bridge supports the optional
I/O and prefetchable memory windows and sets the flag bits in the bridge
resources. This *could* be done once during enumeration except that the
resource allocation code completely clears the flag bits, e.g., in the
pci_assign_unassigned_bridge_resources() path.
The problem with pci_bridge_check_ranges() in the resource allocation path
is that we may allocate resources after devices have been claimed by
drivers, and pci_bridge_check_ranges() *changes* the window registers to
determine whether they're writable. This may break concurrent accesses to
devices behind the bridge.
Add a new pci_read_bridge_windows() to determine whether a bridge supports
the optional windows, call it once during enumeration, remember the
results, and change pci_bridge_check_ranges() so it doesn't touch the
bridge windows but sets the flag bits based on those remembered results.
Signed-off-by: Lei Chen <lennychen@tencent.com>
currently sas disk detected order is decided by the first interrupt
generated order, not by slot number, fix it by slot number.
Signed-off-by: Xiaoming Gao <newtongao@tencent.com>
The commit b3aa14f02254 ("iommu: remove the mapping_error dma_map_ops
method") incorrectly changed the checking from dma_ops_alloc_iova() in
map_sg() causes a crash under memory pressure as dma_ops_alloc_iova()
never return DMA_MAPPING_ERROR on failure but 0, so the error handling
is all wrong.
kernel BUG at drivers/iommu/iova.c:801!
Workqueue: kblockd blk_mq_run_work_fn
RIP: 0010:iova_magazine_free_pfns+0x7d/0xc0
Call Trace:
free_cpu_cached_iovas+0xbd/0x150
alloc_iova_fast+0x8c/0xba
dma_ops_alloc_iova.isra.6+0x65/0xa0
map_sg+0x8c/0x2a0
scsi_dma_map+0xc6/0x160
pqi_aio_submit_io+0x1f6/0x440 [smartpqi]
pqi_scsi_queue_command+0x90c/0xdd0 [smartpqi]
scsi_queue_rq+0x79c/0x1200
blk_mq_dispatch_rq_list+0x4dc/0xb70
blk_mq_sched_dispatch_requests+0x249/0x310
__blk_mq_run_hw_queue+0x128/0x200
blk_mq_run_work_fn+0x27/0x30
process_one_work+0x522/0xa10
worker_thread+0x63/0x5b0
kthread+0x1d2/0x1f0
ret_from_fork+0x22/0x40
Fixes: b3aa14f02254 ("iommu: remove the mapping_error dma_map_ops method")
Signed-off-by: Qian Cai <cai@lca.pw>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
After the conversion to lock-less dma-api call the
increase_address_space() function can be called without any
locking. Multiple CPUs could potentially race for increasing
the address space, leading to invalid domain->mode settings
and invalid page-tables. This has been happening in the wild
under high IO load and memory pressure.
Fix the race by locking this operation. The function is
called infrequently so that this does not introduce
a performance regression in the dma-api path again.
Reported-by: Qian Cai <cai@lca.pw>
Fixes: 256e4621c21a ('iommu/amd: Make use of the generic IOVA allocator')
Signed-off-by: Joerg Roedel <jroedel@suse.de>
This change lowers ring buffer reclaim threshold from 1/2*queue to budget
for better performance. According to our test with qemu + dpdk, packet
dropping happens when the guest is not able to provide free buffer in
avail ring timely with default 1/2*queue. The value in the patch has been
tested and does show better performance.
Test setup: iperf3 to generate packets to guest (total 30mins, pps 400k, UDP)
avg packets drop before: 2842
avg packets drop after: 360(-87.3%)
Further, current code suffers from a starvation problem: the amount of
work done by try_fill_recv is not bounded by the budget parameter, thus
(with large queues) once in a while userspace gets blocked for a long
time while queue is being refilled. Trigger refills earlier to make sure
the amount of work to do is limited.
Signed-off-by: jiangkidd <jiangkidd@hotmail.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
this read can cause IO error when we disconnect nbd with qemu-nbd.
upstream: fe1f9e6659ca6124f500a0f829202c7c902fab0c
signed-off-by: Jia Wang <jasperwang@tencent.com>
Backported from linux-tip ras/core
x86/MCE/AMD, EDAC/mce_amd: Add new MP5, NBIO, and PCIE SMCA bank types
x86/MCE/AMD, EDAC/mce_amd: Add new McaTypes for CS, PSP, and SMU units
x86/MCE/AMD, EDAC/mce_amd: Add new error descriptions for some SMCA bank types
Signed-off-by: Xiaoming Gao <newtongao@tencent.com>
leolingli