Ensure the PV ring is drained on disconnect
If the Xen PV guest VM sends a close whilst there is outstanding I/O being processed that IO needs to be completed and drained before unrealizing the rings or SEGVs will occurr when the I/O does complete and tries to update an already unmapped grant entry.
PATCH 1/1 Ensure the PV ring is drained on disconnect
Also ensure all pending AIO is complete.
--- hw/block/dataplane/xen-block.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+)
hw/block/dataplane/xen-block.c +16 -0
@@ -523,6 +523,10 @@ static bool xen_block_handle_requests(XenBlockDataPlane *dataplane) dataplane->more_work =0 ; +if (dataplane->sring ==0 ) { +return done_something; + } + rc = dataplane->rings.common.req_cons; rp = dataplane->rings.common.sring->req_prod;xen_rmb ();/* Ensure we see queued requests up to 'rp'. */
@@ -666,11 +670,23 @@ void xen_block_dataplane_destroy(XenBlockDataPlane *dataplane)void xen_block_dataplane_stop (XenBlockDataPlane *dataplane) {XenDevice *xendev; +XenBlockRequest *request, *next;if (!dataplane) {return ; } +/* Ensure we have drained the ring */ +do { +xen_block_handle_requests (dataplane); + }while (dataplane->more_work); + +/* Now ensure that all inflight requests are complete */ + QLIST_FOREACH_SAFE(request, &dataplane->inflight, list, next) { +blk_aio_flush (request->dataplane->blk, xen_block_complete_aio, + request); + } + xendev = dataplane->xendev;aio_context_acquire (dataplane->ctx);
Signature · 1 line
2.39.2
cover Updated: Ensure PV ring is drained on disconenct
Updated patch to address intermittent SIGSEGV on domain disconnect/shutdown.
Mark Syms (1):
Ensure the PV ring is drained on disconnect
hw/block/dataplane/xen-block.c | 31 +++++++++++++++++++++++++------ 1 file changed, 25 insertions(+), 6 deletions(-)
Signature · 13 lines
2.40.0 From 21724baa15a72534d98aa2653e9ec39e83559319 Mon Sep 17 00:00:00 2001 From: Mark Syms <[email protected]> Date: Thu, 20 Apr 2023 11:08:34 +0100 Subject: [PATCH 1/1] Ensure the PV ring is drained on disconnect Also ensure all pending AIO is complete. Signed-off-by: Mark Syms <[email protected]> --- hw/block/dataplane/xen-block.c | 31 +++++++++++++++++++++++++------ 1 file changed, 25 insertions(+), 6 deletions(-)
hw/block/dataplane/xen-block.c +25 -6
@@ -523,6 +523,10 @@ static bool xen_block_handle_requests(XenBlockDataPlane *dataplane) dataplane->more_work =0 ; +if (dataplane->sring ==0 ) { +return done_something; + } + rc = dataplane->rings.common.req_cons; rp = dataplane->rings.common.sring->req_prod;xen_rmb ();/* Ensure we see queued requests up to 'rp'. */
@@ -666,14 +670,35 @@ void xen_block_dataplane_destroy(XenBlockDataPlane *dataplane)void xen_block_dataplane_stop (XenBlockDataPlane *dataplane) {XenDevice *xendev; +XenBlockRequest *request, *next;if (!dataplane) {return ; } +/* We're about to drain the ring. We can cancel the scheduling of any +* bottom half now */ +qemu_bh_cancel (dataplane->bh); + +/* Ensure we have drained the ring */ +aio_context_acquire (dataplane->ctx); +do { +xen_block_handle_requests (dataplane); + }while (dataplane->more_work); +aio_context_release (dataplane->ctx); + +/* Now ensure that all inflight requests are complete */ +while (!QLIST_EMPTY (&dataplane->inflight)) { + QLIST_FOREACH_SAFE(request, &dataplane->inflight, list, next) { +blk_aio_flush (request->dataplane->blk, xen_block_complete_aio, + request); + } + } + xendev = dataplane->xendev;aio_context_acquire (dataplane->ctx); +if (dataplane->event_channel) {/* Only reason for failure is a NULL channel */ xen_device_set_event_channel_context(xendev, dataplane->event_channel,
@@ -684,12 +709,6 @@ void xen_block_dataplane_stop(XenBlockDataPlane *dataplane)blk_set_aio_context (dataplane->blk,qemu_get_aio_context (), &error_abort);aio_context_release (dataplane->ctx); -/* -* Now that the context has been moved onto the main thread, cancel -* further processing. -*/ -qemu_bh_cancel (dataplane->bh); -if (dataplane->event_channel) {Error *local_err =NULL ;
Signature · 1 line
2.40.0
Re: [PATCH 0/1] Updated: Ensure PV ring is drained on disconenct
Quoted 17 lines · Mark Syms
Updated patch to address intermittent SIGSEGV on domain disconnect/shutdown.
Mark Syms (1):
Ensure the PV ring is drained on disconnecthw/block/dataplane/xen-block.c | 31 +++++++++++++++++++++++++------ 1 file changed, 25 insertions(+), 6 deletions(-)Signature · 1 line
2.40.0Quoted 1 line
From
21724baa15a72534d98aa2653e9ec39e83559319Mon Sep 17 00:00:00 2001From: Mark Syms <[email protected]>
Date: Thu, 20 Apr 2023 11:08:34 +0100
Subject: [PATCH 1/1] Ensure the PV ring is drained on disconnectAlso ensure all pending AIO is complete.
Hi Mark, can you please add more info on the problem you are trying to solve? Also add any stacktrace if you get any due to this error.
Quoted 10 lines
Signed-off-by Mark Syms <[email protected]>--- hw/block/dataplane/xen-block.c | 31 +++++++++++++++++++++++++------ 1 file changed, 25 insertions(+), 6 deletions(-)hw/block/dataplane/xen-block.c +0
-0
hw/block/dataplane/xen-block.c static bool xen_block_handle_requests(XenBlockDataPlane *dataplane) · line 527
Quoted 5 lines
hw/block/dataplane/xen-block.c +2
-0@@ -523,6 +523,10 @@ static bool xen_block_handle_requests(XenBlockDataPlane *dataplane) dataplane->more_work =0 ; +if (dataplane->sring ==0 ) { +return done_something;
done_something cannot be changed by now, so I would just do
return false;
Quoted 11 lines
hw/block/dataplane/xen-block.c +3
-0+ } + rc = dataplane->rings.common.req_cons; rp = dataplane->rings.common.sring->req_prod;xen_rmb ();/* Ensure we see queued requests up to 'rp'. */ @@ -666,14 +670,35 @@ void xen_block_dataplane_destroy(XenBlockDataPlane *dataplane)void xen_block_dataplane_stop (XenBlockDataPlane *dataplane) {XenDevice *xendev; +XenBlockRequest *request, *next;
hw/block/dataplane/xen-block.c void xen_block_dataplane_destroy(XenBlockDataPlane *dataplane) · line 684
Quoted 10 lines
hw/block/dataplane/xen-block.c +6
-0if (!dataplane) {return ; } +/* We're about to drain the ring. We can cancel the scheduling of any +* bottom half now */ +qemu_bh_cancel (dataplane->bh); + +/* Ensure we have drained the ring */ +aio_context_acquire (dataplane->ctx);
Would it make sense to move the 2 loops below under the existing aio_context_acquire also below?
hw/block/dataplane/xen-block.c void xen_block_dataplane_destroy(XenBlockDataPlane *dataplane) · line 696
Quoted 12 lines
hw/block/dataplane/xen-block.c +12
-0+do { +xen_block_handle_requests (dataplane); + }while (dataplane->more_work); +aio_context_release (dataplane->ctx); + +/* Now ensure that all inflight requests are complete */ +while (!QLIST_EMPTY (&dataplane->inflight)) { + QLIST_FOREACH_SAFE(request, &dataplane->inflight, list, next) { +blk_aio_flush (request->dataplane->blk, xen_block_complete_aio, + request); + } + }
especially because I would think that blk_aio_flush needs to be called with aio_context_acquired ?
hw/block/dataplane/xen-block.c void xen_block_dataplane_destroy(XenBlockDataPlane *dataplane) · line 701
Quoted 4 lines
hw/block/dataplane/xen-block.c +1
-0xendev = dataplane->xendev;aio_context_acquire (dataplane->ctx); +
move the new code here
Quoted 19 lines
hw/block/dataplane/xen-block.c +0
-6if (dataplane->event_channel) {/* Only reason for failure is a NULL channel */ xen_device_set_event_channel_context(xendev, dataplane->event_channel,@@ -684,12 +709,6 @@ void xen_block_dataplane_stop(XenBlockDataPlane *dataplane)blk_set_aio_context (dataplane->blk,qemu_get_aio_context (), &error_abort);aio_context_release (dataplane->ctx); -/* -* Now that the context has been moved onto the main thread, cancel -* further processing. -*/ -qemu_bh_cancel (dataplane->bh); -if (dataplane->event_channel) {Error *local_err =NULL ;Signature · 1 line
2.40.0