Fix RBM_ZERO_AND_LOCK mode to not acquire lock on local buffers.
authorHeikki Linnakangas <heikki.linnakangas@iki.fi>
Wed, 13 May 2015 06:44:43 +0000 (09:44 +0300)
committerHeikki Linnakangas <heikki.linnakangas@iki.fi>
Wed, 13 May 2015 07:06:47 +0000 (10:06 +0300)
commitf6c4a8690c9374be828c7fca4c2b6199f39b27fa
tree03405cf9f77f6b9a65751c8468427aa67fde93d1
parentdd75518d523a1e3650b5d5ad20c755a000425739
Fix RBM_ZERO_AND_LOCK mode to not acquire lock on local buffers.

Commit 81c45081 introduced a new RBM_ZERO_AND_LOCK mode to ReadBuffer, which
takes a lock on the buffer before zeroing it. However, you cannot take a
lock on a local buffer, and you got a segfault instead. The version of that
patch committed to master included a check for !isLocalBuf, and therefore
didn't crash, but oddly I missed that in the back-patched versions. This
patch adds that check to the back-branches too.

RBM_ZERO_AND_LOCK mode is only used during WAL replay, and in hash indexes.
WAL replay only deals with shared buffers, so the only way to trigger the
bug is with a temporary hash index.

Reported by Artem Ignatyev, analysis by Tom Lane.
src/backend/storage/buffer/bufmgr.c