RepoMirrors
/
ceph
mirror of https://github.com/ceph/ceph
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

crimson/osd/pg: fix dangerous lambda captures 

In PG::submit_transaction(), osd_op_p was first moved to
ReplicatedBackend::submit_transaction(), and later its at_version is
captured in another lambda. The latter may be already empty at the time
of the capture.

Signed-off-by: Xuehan Xu <xuxuehan@qianxin.com>
This commit is contained in:
Xuehan Xu 2024-11-02 10:01:52 +08:00
parent 1c93566522
commit cbd22c7ba2
1 changed files with 4 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
src/crimson/osd/pg.cc
View File

@ -924,8 +924,9 @@ PG::submit_transaction(
}
epoch_t map_epoch = get_osdmap_epoch();
auto at_version = osd_op_p.at_version;
peering_state.pre_submit_op(obc->obs.oi.soid, log_entries, osd_op_p.at_version);
peering_state.pre_submit_op(obc->obs.oi.soid, log_entries, at_version);
peering_state.update_trim_to();
ceph_assert(!log_entries.empty());
@ -947,8 +948,8 @@ PG::submit_transaction(
co_return std::make_tuple(
std::move(submitted),
all_completed.then_interruptible(
[this, last_complete=peering_state.get_info().last_complete,
at_version=osd_op_p.at_version](auto acked) {
[this, at_version,
last_complete=peering_state.get_info().last_complete](auto acked) {
for (const auto& peer : acked) {
peering_state.update_peer_last_complete_ondisk(
peer.shard, peer.last_complete_ondisk);