Veristat memory accounting #9050
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
eddyz87
force-pushed
the
veristat-memory-accounting
branch
3 times, most recently
from
d0433ea to
411acf6
Compare
kernel-patches-daemon-bpf
bot
force-pushed
the
bpf-next_base
branch
7 times, most recently
from
76357bc to
0f8e0ae
Compare
This commit adds __GFP_ACCOUNT flag to verifier induced memory allocations. The intent is to account for all allocations reachable from BPF_PROG_LOAD command, which is needed to track verifier memory consumption in veristat. This includes allocations done in verifier.c, and some allocations in btf.c, functions in log.c do not allocate. There is also a utility function bpf_memcg_flags() which selectively adds GFP_ACCOUNT flag depending on the `cgroup.memory=nobpf` option. As far as I understand [1], the idea is to remove bpf_prog instances and maps from memcg accounting as these objects do not strictly belong to cgroup, hence it should not apply here. (btf_parse_fields() is reachable from both program load and map creation, but allocated record is not persistent as is freed as soon as map_check_btf() exits). [1] https://lore.kernel.org/all/[email protected]/ Signed-off-by: Eduard Zingerman <[email protected]>
This commit adds a new field mem_peak / "Peak memory" field to a set
of gathered statistics. The field is intended as an estimate for peak
verifier memory consumption for processing of a given program.
Mechanically stat is collected as follows:
- At the beginning of handle_verif_mode() a new cgroup namespace is
created and cgroup fs is mounted in this namespace, memory
controller is enabled for the root cgroup.
- At each program load:
- bpf_object__load() is split into bpf_object__prepare() and
bpf_object__load() to avoid accounting for memory allocated for
maps;
- before bpf_object__load() a new cgroup is created and veristat
process enters this cgroup, "memory.peak" of the new cgroup is
stashed;
- after bpf_object__load() the difference between current
"memory.peak" and stashed "memory.peak" is used as a metric,
veristat exits the cgroup and cgroup is discarded.
If any of the above steps fails veristat would proceed w/o collecting
mem_peak information for a program.
The change has impact on veristat running time, e.g. for all
test_progs object files there is an increase from 82s to 102s.
I take a correlation between "Peak states" and "Peak memory" fields as
a sanity check for gathered statistics, e.g. here is a sample of data
for sched_ext programs:
File Program Peak states Peak memory (KiB)
--------- -------------------- ----------- -----------------
bpf.bpf.o lavd_select_cpu 1311 26256
bpf.bpf.o lavd_enqueue 1140 22720
bpf.bpf.o layered_enqueue 777 11504
bpf.bpf.o layered_dispatch 578 7976
bpf.bpf.o lavd_dispatch 634 6204
bpf.bpf.o rusty_init 343 5352
bpf.bpf.o lavd_init 361 5092
...
bpf.bpf.o rusty_exit_task 36 256
bpf.bpf.o rusty_running 19 256
bpf.bpf.o bpfland_dispatch 3 0
bpf.bpf.o bpfland_enable 1 0
Signed-off-by: Eduard Zingerman <[email protected]>
eddyz87
force-pushed
the
veristat-memory-accounting
branch
from
411acf6 to
a705c27
Compare
kernel-patches-daemon-bpf
bot
force-pushed
the
bpf-next_base
branch
17 times, most recently
from
dd96e0b to
9860b87
Compare
kernel-patches-daemon-bpf
bot
force-pushed
the
bpf-next_base
branch
from
9860b87 to
7801208
Compare
|
Automatically cleaning up stale PR; feel free to reopen if needed |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.