Show multiple user PF's in the output#3
Show multiple user PF's in the output#3damiandcoherent wants to merge 1 commit intoDispatchCode:mainfrom
Conversation
damiandcoherent
force-pushed
the
main
branch
from
1db8670 to
44c54e4
Compare
sigsegv-monitor.bpf.c
Outdated
| event->cr2_tai[i] = stat->tai; | ||
|
|
||
| ++event->cr2_userpf_entry_count; | ||
| } |
There was a problem hiding this comment.
Should we perhaps break the loop if stat == NULL?
There was a problem hiding this comment.
This should never be the case, it is there just for the verifier. If something horribly wrong happens, shouldn't we be printing as much as we can?
There was a problem hiding this comment.
I think my point was that the ring buffer get function returns NULL if you're past the write end (and the buffer is not full yet). Unfortunately, I don't know how to view my comment on the original code in GitHub :/
| cr2 = ctx->address; | ||
| stat.cr2 = ctx->address; | ||
| stat.err = ctx->error_code; | ||
| stat.tai = bpf_ktime_get_tai_ns(); |
There was a problem hiding this comment.
Have you measured the performance impact when executing lots of page faults?
There was a problem hiding this comment.
Yes, I did not see any drastic change in the performance. Time per page fault appears to be in the range 0.75-1.1µs
damiandcoherent
force-pushed
the
main
branch
2 times, most recently
from
9e2dde9 to
adb47a7
Compare
damiandcoherent
force-pushed
the
main
branch
4 times, most recently
from
2c96b30 to
dbd19c0
Compare
damiandcoherent
force-pushed
the
main
branch
from
dbd19c0 to
6c07d5c
Compare
3 participants
No description provided.