perf: Reduce stack usage (and misc bits)
Status from this message: The following commit has been merged into the perf/urgent branch of tip:
Get series with b4
b4 am [email protected]Versions newest first
Patches in v1 Applied
| # | Subject and files | Size | Replies | Applied |
|---|---|---|---|---|
| 1/6 | perf: Reduce stack usage of perf_output_begin()arch/ arch/ arch/ +3 more | +37 | 0 replies | |
| 2/6 | perf/x86: Reduce stack usage for x86_pmu::drain_pebs()arch/ arch/ arch/ | +27 | 0 replies | |
| 3/6 | perf: Fix get_recursion_context()kernel/ | +1 | 0 replies | |
| 4/6 | perf: Optimize get_recursion_context()kernel/ | +8 | 10 replies | |
| 5/6 | perf/arch: Remove perf_sample_data::regs_user_copyarch/ arch/ arch/ +7 more | +22 | 0 replies | |
| 6/6 | perf/x86: Make dummy_iregs staticarch/ | +1 | 0 replies |
Reviews trailers per patch
1 of 6 patches has a Reviewed-by or Acked-by.
| Patch | Steven Rostedt | Jesper Dangaard Brouer |
|---|---|---|
| 1/6 | ||
| 2/6 | ||
| 3/6 | ||
| 4/6 | A-b | |
| 5/6 | T-b | |
| 6/6 |
Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.
Cover letter [PATCH 0/6] perf: Reduce stack usage (and misc bits)
Hi,
Steve reported an NMI stack overflow when running perf and function tracing together. Thomas found that we had 4 copies of struct perf_sample_data on-stack.
These here patches reduce that to 2 copies and half the size of it.
So just for perf_sample_data we go from 4*384=1536 to 2*192=384 bytes of stack. Also remove one struct pt_regs instance from __intel_pmu_pebs_event(); it has another instance in struct x86_perf_regs which I haven't yet managed to offload.
Perf seems to still work... :-)