tjh.dev / kernel
Linux kernel mirror (for testing) git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel os linux
1
fork

Configure Feed

Select the types of activity you want to include in your feed.

drm/xe/gsc: Fix GSC proxy cleanup on early initialization failure

xe_gsc_proxy_remove undoes what is done in both xe_gsc_proxy_init and
xe_gsc_proxy_start; however, if we fail between those 2 calls, it is
possible that the HW forcewake access hasn't been initialized yet and so
we hit errors when the cleanup code tries to write GSC register. To
avoid that, split the cleanup in 2 functions so that the HW cleanup is
only called if the HW setup was completed successfully.

Since the HW cleanup (interrupt disabling) is now removed from
xe_gsc_proxy_remove, the cleanup on error paths in xe_gsc_proxy_start
must be updated to disable interrupts before returning.

Fixes: ff6cd29b690b ("drm/xe: Cleanup unwind of gt initialization")
Signed-off-by: Zhanjun Dong <zhanjun.dong@intel.com>
Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Link: https://patch.msgid.link/20260220225308.101469-1-zhanjun.dong@intel.com

authored by

Zhanjun Dong and committed by
Daniele Ceraolo Spurio 2b37c401 c692ae39

+37 -8
+35 -8
drivers/gpu/drm/xe/xe_gsc_proxy.c
··· 435 435 return 0; 436 436 } 437 437 438 - static void xe_gsc_proxy_remove(void *arg) 438 + static void xe_gsc_proxy_stop(struct xe_gsc *gsc) 439 439 { 440 - struct xe_gsc *gsc = arg; 441 440 struct xe_gt *gt = gsc_to_gt(gsc); 442 441 struct xe_device *xe = gt_to_xe(gt); 443 - 444 - if (!gsc->proxy.component_added) 445 - return; 446 442 447 443 /* disable HECI2 IRQs */ 448 444 scoped_guard(xe_pm_runtime, xe) { ··· 451 455 } 452 456 453 457 xe_gsc_wait_for_worker_completion(gsc); 458 + gsc->proxy.started = false; 459 + } 460 + 461 + static void xe_gsc_proxy_remove(void *arg) 462 + { 463 + struct xe_gsc *gsc = arg; 464 + struct xe_gt *gt = gsc_to_gt(gsc); 465 + struct xe_device *xe = gt_to_xe(gt); 466 + 467 + if (!gsc->proxy.component_added) 468 + return; 469 + 470 + /* 471 + * GSC proxy start is an async process that can be ongoing during 472 + * Xe module load/unload. Using devm managed action to register 473 + * xe_gsc_proxy_stop could cause issues if Xe module unload has 474 + * already started when the action is registered, potentially leading 475 + * to the cleanup being called at the wrong time. Therefore, instead 476 + * of registering a separate devm action to undo what is done in 477 + * proxy start, we call it from here, but only if the start has 478 + * completed successfully (tracked with the 'started' flag). 479 + */ 480 + if (gsc->proxy.started) 481 + xe_gsc_proxy_stop(gsc); 454 482 455 483 component_del(xe->drm.dev, &xe_gsc_proxy_component_ops); 456 484 gsc->proxy.component_added = false; ··· 530 510 */ 531 511 int xe_gsc_proxy_start(struct xe_gsc *gsc) 532 512 { 513 + struct xe_gt *gt = gsc_to_gt(gsc); 533 514 int err; 534 515 535 516 /* enable the proxy interrupt in the GSC shim layer */ ··· 542 521 */ 543 522 err = xe_gsc_proxy_request_handler(gsc); 544 523 if (err) 545 - return err; 524 + goto err_irq_disable; 546 525 547 526 if (!xe_gsc_proxy_init_done(gsc)) { 548 - xe_gt_err(gsc_to_gt(gsc), "GSC FW reports proxy init not completed\n"); 549 - return -EIO; 527 + xe_gt_err(gt, "GSC FW reports proxy init not completed\n"); 528 + err = -EIO; 529 + goto err_irq_disable; 550 530 } 551 531 532 + gsc->proxy.started = true; 552 533 return 0; 534 + 535 + err_irq_disable: 536 + gsc_proxy_irq_toggle(gsc, false); 537 + return err; 553 538 }
+2
drivers/gpu/drm/xe/xe_gsc_types.h
··· 58 58 struct mutex mutex; 59 59 /** @proxy.component_added: whether the component has been added */ 60 60 bool component_added; 61 + /** @proxy.started: whether the proxy has been started */ 62 + bool started; 61 63 /** @proxy.bo: object to store message to and from the GSC */ 62 64 struct xe_bo *bo; 63 65 /** @proxy.to_gsc: map of the memory used to send messages to the GSC */