Hi Shiraz & Jason,
We have observed a crash when run perftest on a hisilicon arm64 platform in kernel-5.2-rc4.
We also tested with different kernel version and found it started from the the following commit:
d10bcf947a3e ("RDMA/umem: Combine contiguous PAGE_SIZE regions in SGEs")
Could you please share any hint how to resolve this kind issue?
Thanks!
The command we are using to run the perftest is as:
ib_write_bw -d hns_2 -n 5 &
ib_write_bw -d hns_2 -n 5 192.168.10.110
The log is as below:
root@(none)$ ib_write_bw -d hns_2 -n 5 &
[[1] 1149
484.818219] BUG: Bad rss-counter state mm:(____ptrval____) idx:0 val:-1
[ 484.818219] BUG: Bad rss-counter state mm:(____ptrval____) idx:0 val:-1
[ 484.832330] BUG: Bad rss-counter state mm:(____ptrval____) idx:1 val:1
[ 484.832330] BUG: Bad rss-counter state mm:(____ptrval____) idx:1 val:1
************************************
* Waiting for client to connect... *
************************************
root@(none)$ ib_write_bw -d hns_2 -n 5 192.168.10.110
[ 488.215284] BUG: Bad rss-counter state mm:(____ptrval____) idx:0 val:-1
[ 488.215284] BUG: Bad rss-counter state mm:(____ptrval____) idx:0 val:-1
[ 488.228526] BUG: Bad rss-counter state mm:(____ptrval____) idx:1 val:1
[ 488.228526] BUG: Bad rss-counter state mm:(____ptrval____) idx:1 val:1
cqe = 5, less than minimum CQE nu[mber.
cqe = 5, less than minimum CQE number.
---------------------------------------------------------------------------------------
RDMA_Write BW Test
Dual-port : OFF Device : hns_2
Number of qps : 1 Transport type : IB
Connection type : RC Using SRQ : OFF
TX depth : 5
CQ Moderation : 5
Mtu : 1024[B]
Link type : Ethernet
GID index : 2
Max inline data : 0[B]
rdma_cm QPs : OFF
Data ex. method : Ethernet
---------------------------------------------------------------------------------------
local address: LID 0000 QPN 0x000d PSN 0x7d8986 RKey 0x000200 VAddr 0x00ffffbd254000
GID: 00:00:00:00:00:00:00:00:00:00:255:255:192:168:10:110
489.244399] BUG: Bad page map in process ib_write_bw pte:e800305b55ef53 pmd:3fb6044003
[ 489.244399] BUG: Bad page map in process ib_write_bw pte:e800305b55ef53 pmd:3fb6044003
[ 489.327926] page:ffff7e1d416d5780 refcount:1 mapcount:-1 mapping:ffff875fb60453c9 index:0x0
[ 489.327926] page:ffff7e1d416d5780 refcount:1 mapcount:-1 mapping:ffff875fb60453c9 index:0x0
[ 489.344651] anon
[ 489.344651] anon
[ 489.344652] flags: 0x5fffe00000080034(uptodate|lru|active|swapbacked)
[ 489.344652] flags: 0x5fffe00000080034(uptodate|lru|active|swapbacked)
[ 489.361380] raw: 5fffe00000080034 ffff7e1d416d5a48 ffff7e1d416d5c48 ffff875fb60453c9
[ 489.361380] raw: 5fffe00000080034 ffff7e1d416d5a48 ffff7e1d416d5c48 ffff875fb60453c9
[ 489.376884] raw: 0000000000000000 0000000000000000 00000001fffffffe ffff874fbf6b4000
[ 489.376884] raw: 0000000000000000 0000000000000000 00000001fffffffe ffff874fbf6b4000
[ 489.392387] page dumped because: bad pte
[ 489.392387] page dumped because: bad pte
[ 489.400228] page->mem_cgroup:ffff874fbf6b4000
[ 489.400228] page->mem_cgroup:ffff874fbf6b4000
[ 489.408941] addr:(____ptrval____) vm_flags:00100073 anon_vma:(____ptrval____) mapping:(____ptrval____) index:ffffb5a54
[ 489.408941] addr:(____ptrval____) vm_flags:00100073 anon_vma:(____ptrval____) mapping:(____ptrval____) index:ffffb5a54
Best Regards,
Lijun
