From a478f0b418bd3dd5fbcc6260979865160ca25cf0 Mon Sep 17 00:00:00 2001 From: Arthur Hanson Rasmusson Date: Mon, 9 Nov 2026 19:27:48 -0500 Subject: vgsp: raise the GSP engine event interrupt so guest ISR drains events and wakes poll() waiters The emulated GSP only signalled the MSI-X eventfd when it queued a status record. The guest's primary ISR (nvidia_isr) decides pending from the CPU_INTR leaf registers, which we never set for the GSP stall vector, so it returned IRQ_NONE: IRQ 33 counted thousands of interrupts while the nvidia event thread sat idle and a client blocked in poll() on /dev/nvidia0 (a GLX swap) waited forever even though the event was already in the status queue. Mirror what a real GSP-RM does when it queues an event: - program RISC-V IRQMASK/IRQDEST for SWGEN0 (read-only to the guest; the GSP firmware owns them), - set SWGEN0 in the falcon IRQSTAT, - latch GSP stall vector 155 in the CPU interrupt tree (LEAF[4] bit 27, UVM_SHARED subtree 2 -> TOP[0] bit 2), - signal MSI-X. Raise it only for POST_EVENT records (replies are polled by the client). Verified on r1680: the guest now writes IRQSCLR SWGEN0, kgspService_TU102 drains events, and glxgears renders (was a hard hang). --- kernel/gsp_rm_emulation.c | 14 +++++++++----- 1 file changed, 9 insertions(+), 5 deletions(-) (limited to 'kernel/gsp_rm_emulation.c') diff --git a/kernel/gsp_rm_emulation.c b/kernel/gsp_rm_emulation.c index c14a767..f395ce6 100644 --- a/kernel/gsp_rm_emulation.c +++ b/kernel/gsp_rm_emulation.c @@ -737,11 +737,15 @@ static int gsp_rm_send_flags(struct gsp_emu_context *ctx, u32 function, return ret; } - /* Notify the guest (MSI-X vector 0). The driver's bottom-half drains - * events under the GPU lock; the rpcRecvPoll poller holds the same - * lock from send to poll, so a reply can't be consumed by the event - * drain while a poller waits for it (same as real GSP-RM). */ - gsp_emu_inject_completion_interrupt(ctx, rm->stat_seq); + /* Notify the guest. An event must also raise the GSP engine interrupt + * (falcon SWGEN0 + CPU_INTR leaf/top) so the guest's ISR drains the + * status queue and nv_post_event() wakes a client blocked in poll() on + * /dev/nvidia0 (e.g. a GLX swap). Replies only need the MSI-X ping + * because the client polls the queue for them. */ + if (function == GSPRM_RPC_EVENT_POST_EVENT) + gsp_emu_raise_gsp_event_irq(ctx); + else + gsp_emu_inject_completion_interrupt(ctx, rm->stat_seq); mutex_unlock(&rm->stat_lock); return 0; -- cgit v1.2.3