From 55ffe494a749cf0ebf10b3146b8c59a4ab81aa5f Mon Sep 17 00:00:00 2001 From: Arthur Hanson Rasmusson Date: Mon, 21 Sep 2026 21:08:06 -0400 Subject: kernel: TLB invalidate auto-complete (MMU_INVALIDATE trigger) + fn-10 queue logging - NV_VIRTUAL_FUNCTION_PRIV_MMU_INVALIDATE (BAR0+0x30B0, TRIGGER=bit31): driver sets TRIGGER then spins in kgmmuCheckPendingInvalidates_TU102 polling for it to clear. Our shadow held TRIGGER forever -> nv_open_q hard-wedged in RmInitAdapter (TLB invalidate during dmaAllocMapping). Auto-clear the trigger bit in gsp_emu_poll_boot (instant completion). - fn-10/47 handler logs hObjectOld + cmd_read/stat_write/drv_rp/stat_seq. Confirmed: drv_rp tracks stat_write 1:1 (driver drains cleanly, no lapping); frees complete; stat_seq increments per record. - Health-check/crashcat loop (kgspHealthCheck -> crashcatEngineGetNext CrashReport never returns NULL) is a SECONDARY symptom triggered by an RPC failure during RmInitAdapter - under investigation. --- kernel/gsp_rm_emulation.c | 20 ++++++++++++-------- 1 file changed, 12 insertions(+), 8 deletions(-) (limited to 'kernel/gsp_rm_emulation.c') diff --git a/kernel/gsp_rm_emulation.c b/kernel/gsp_rm_emulation.c index 0a040d5..67115b3 100644 --- a/kernel/gsp_rm_emulation.c +++ b/kernel/gsp_rm_emulation.c @@ -1289,18 +1289,22 @@ static void gsp_rm_handle_element(struct gsp_emu_context *ctx, const u8 *elem) * driver's rpcRecvPoll completes. */ rm->unknown_rpcs++; if (function == 10 || function == 47) { - /* FREE/UNLOADING: log hObject for unload-cycle debugging */ + /* FREE/UNLOADING: log hObject + queue positions for + * unload-cycle debugging. driver_rp = the peer's status + * read pointer (swap-rx): if it lags stat_write, the + * driver isn't draining our replies. */ const u8 *p = elem + GSPRM_ELEM_HDR_SIZE + sizeof(struct gsprm_rpc_header); - u32 hc = 0, ho = 0; + u32 hc = 0, ho = 0, driver_rp = 0; - if (hdr->length >= sizeof(*hdr) + 8) { - memcpy(&hc, p, 4); - memcpy(&ho, p + 4, 4); - } - nv_vgpu_info(ctx->vgpu, "GSPRM: RPC %s(%u) seq=%u hClient=0x%x hObject=0x%x cmd_read=%u", + if (hdr->length >= sizeof(*hdr) + 12) + memcpy(&ho, p + 8, 4); /* hObjectOld */ + gsp_rm_read_buf(ctx, rm->cmd_off + 0x20, + &driver_rp, sizeof(driver_rp)); + nv_vgpu_info(ctx->vgpu, "GSPRM: RPC %s(%u) seq=%u hRoot=0x%x hObjOld=0x%x cmd_read=%u stat_write=%u drv_rp=%u stat_seq=%u", gsp_rm_rpc_name(function), function, sequence, - hc, ho, rm->cmd_read); + hc, ho, rm->cmd_read, rm->stat_write, + driver_rp, rm->stat_seq); } else { nv_vgpu_info(ctx->vgpu, "GSPRM: RPC %s(%u) seq=%u len=%u -> generic SUCCESS reply (unknown #%llu)", gsp_rm_rpc_name(function), function, sequence, -- cgit v1.2.3