Merge branch 'ibmveth-fix-open-fail-unwind-after-lan-registration'

Mingming Cao says:

====================
ibmveth: fix open-fail unwind after LAN registration

ibmveth_open() has two independent unwind holes after the logical LAN
is set up. Neither needs the MQ RX series. This posting is against
current net.git.

Patch 1 issues h_free_logical_lan() on every post-register failure.
A pool allocation failure jumped to out_free_buffer_pools without
the hypercall, so PHYP still owned the buffer-list page when it was
unmapped. request_irq() failure already issued H_FREE before taking
the same label.

  Fixes: d43732ce02 ("ibmveth: properly unwind on init errors")

Patch 2 gives TX LTBs their own walk. out_free_buffer_pools reuses
open()'s loop index, so after the pool unwind i is -1 and the TX
buffers leak. request_irq() failure has the same leak. The TX-fail
goto on that walk also skips dma_unmap of the filter list
(pre-existing since the LTB was added).

  Fixes: d926793c1d ("ibmveth: Implement multi queue on xmit")

The MQ RX series on net-next keeps the helper versions of these
guards and does not depend on this pair. If both land, keep the
helpers in that series; these two patches are the current
single-queue open-fail path only.
====================

Link: https://patch.msgid.link/cover.1790357373.git.mmc@linux.ibm.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This commit is contained in:
Jakub Kicinski 2026-10-01 15:59:29 -07:00
commit 8e9cb39859