From b7698b917abcbe673cbffc15ee41d95d3daa4af3 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=C5=81ukasz=20Gieryk?= Date: Mon, 9 May 2022 16:16:19 +0200 Subject: hw/nvme: Update the initalization place for the AER queue MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This patch updates the initialization place for the AER queue, so it’s initialized once, at controller initialization, and not every time controller is enabled. While the original version works for a non-SR-IOV device, as it’s hard to interact with the controller if it’s not enabled, the multiple reinitialization is not necessarily correct. With the SR/IOV feature enabled a segfault can happen: a VF can have its controller disabled, while a namespace can still be attached to the controller through the parent PF. An event generated in such case ends up on an uninitialized queue. While it’s an interesting question whether a VF should support AER in the first place, I don’t think it must be answered today. Signed-off-by: Łukasz Gieryk Reviewed-by: Klaus Jensen Acked-by: Michael S. Tsirkin Signed-off-by: Klaus Jensen --- hw/nvme/ctrl.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'hw/nvme') diff --git a/hw/nvme/ctrl.c b/hw/nvme/ctrl.c index 20f1a73995..658584d417 100644 --- a/hw/nvme/ctrl.c +++ b/hw/nvme/ctrl.c @@ -6328,8 +6328,6 @@ static int nvme_start_ctrl(NvmeCtrl *n) nvme_set_timestamp(n, 0ULL); - QTAILQ_INIT(&n->aer_queue); - nvme_select_iocs(n); return 0; @@ -6989,6 +6987,7 @@ static void nvme_init_state(NvmeCtrl *n) n->features.temp_thresh_hi = NVME_TEMPERATURE_WARNING; n->starttime_ms = qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL); n->aer_reqs = g_new0(NvmeRequest *, n->params.aerl + 1); + QTAILQ_INIT(&n->aer_queue); list->numcntl = cpu_to_le16(max_vfs); for (i = 0; i < max_vfs; i++) { -- cgit v1.2.3