GFS2: Fix allocation error bug with recursive rgrp glocking

Before this patch function gfs2_write_begin, upon discovering an
error, called gfs2_trim_blocks while the rgrp glock was still held.
That's because gfs2_inplace_release is not called until later.
This patch reorganizes the logic a bit so gfs2_inplace_release
is called to release the lock prior to the call to gfs2_trim_blocks,
thus preventing the glock recursion.

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
This commit is contained in:
Andreas Gruenbacher 2018-05-31 07:20:55 -05:00 committed by Bob Peterson
parent 07e23d68f6
commit 7b5747f43f

View File

@ -747,18 +747,21 @@ static int gfs2_write_begin(struct file *file, struct address_space *mapping,
put_page(page); put_page(page);
gfs2_trans_end(sdp); gfs2_trans_end(sdp);
if (alloc_required) {
gfs2_inplace_release(ip);
if (pos + len > ip->i_inode.i_size) if (pos + len > ip->i_inode.i_size)
gfs2_trim_blocks(&ip->i_inode); gfs2_trim_blocks(&ip->i_inode);
goto out_trans_fail; }
goto out_qunlock;
out_endtrans: out_endtrans:
gfs2_trans_end(sdp); gfs2_trans_end(sdp);
out_trans_fail: out_trans_fail:
if (alloc_required) { if (alloc_required)
gfs2_inplace_release(ip); gfs2_inplace_release(ip);
out_qunlock: out_qunlock:
if (alloc_required)
gfs2_quota_unlock(ip); gfs2_quota_unlock(ip);
}
out_unlock: out_unlock:
if (&ip->i_inode == sdp->sd_rindex) { if (&ip->i_inode == sdp->sd_rindex) {
gfs2_glock_dq(&m_ip->i_gh); gfs2_glock_dq(&m_ip->i_gh);