From: Claude Code Review Bot <claude-review@example.com>
To: dri-devel-reviews@example.com
Subject: Claude review: drm/xe: Handle PUNIT errors by requesting cold-reset recovery
Date: Sun, 12 Apr 2026 14:32:34 +1000 [thread overview]
Message-ID: <review-patch4-20260406142325.157035-10-mallesh.koujalagi@intel.com> (raw)
In-Reply-To: <20260406142325.157035-10-mallesh.koujalagi@intel.com>
Patch Review
**`xe_punit_error_handler()` double-wedging concern:**
```c
void xe_punit_error_handler(struct xe_device *xe)
{
xe_err(xe, "Recovery: Device cold reset required\n");
xe_device_set_wedged_method(xe, DRM_WEDGE_RECOVERY_COLD_RESET);
xe_device_declare_wedged(xe);
}
```
This is called from `handle_soc_internal_errors()`, which is called from `xe_ras_process_errors()`, which is called from `xe_pci_error_mmio_enabled()`. But looking at `xe_pci_error_detected()` for the `pci_channel_io_frozen` case, `xe_pci_error_handling()` already sets `atomic_set(&xe->wedged.flag, 1)` and calls `xe_gt_declare_wedged()`. So by the time `xe_punit_error_handler()` runs and calls `xe_device_declare_wedged()`, the device is already partially wedged. `xe_device_declare_wedged()` uses `atomic_xchg(&xe->wedged.flag, 1)` which will see it's already set and skip the PM runtime get and error message, but it will still call `xe_gt_declare_wedged()` again for all GTs (double wedge). The uevent will still fire (good), but the flow is fragile and depends on ordering assumptions.
For the `pci_channel_io_normal` path (which returns `CAN_RECOVER` and leads to `mmio_enabled`), the wedged flag is NOT pre-set, so `xe_punit_error_handler()` works correctly. But this means the behavior differs depending on which PCI channel state triggered the error — this should be documented.
**Exported function placement:**
`xe_punit_error_handler()` is declared in `xe_ras.h` but it's a fairly high-level operation (wedge + uevent). It might make more sense in `xe_device.c` or at least the name should reflect that it's specific to the RAS/PCI error recovery path, not a general-purpose API.
**Header include ordering in xe_ras.c:**
```c
+#include "xe_device.h"
```
Added between `xe_device_types.h` and `xe_printk.h` — the includes should be in alphabetical order. `xe_device.h` should come before `xe_device_types.h`.
---
Generated by Claude Code Patch Reviewer
next prev parent reply other threads:[~2026-04-12 4:32 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-04-06 14:23 [PATCH v3 0/4] Introduce cold reset recovery method Mallesh Koujalagi
2026-04-06 14:23 ` [PATCH v3 1/4] Introduce Xe Uncorrectable Error Handling Mallesh Koujalagi
2026-04-12 4:32 ` Claude review: " Claude Code Review Bot
2026-04-06 14:23 ` [PATCH v3 2/4] drm: Add DRM_WEDGE_RECOVERY_COLD_RESET recovery method Mallesh Koujalagi
2026-04-08 7:46 ` Raag Jadav
2026-04-12 4:32 ` Claude review: " Claude Code Review Bot
2026-04-06 14:23 ` [PATCH v3 3/4] drm/doc: Document " Mallesh Koujalagi
2026-04-08 8:01 ` Raag Jadav
2026-04-12 4:32 ` Claude review: " Claude Code Review Bot
2026-04-06 14:23 ` [PATCH v3 4/4] drm/xe: Handle PUNIT errors by requesting cold-reset recovery Mallesh Koujalagi
2026-04-08 8:09 ` Raag Jadav
2026-04-12 4:32 ` Claude Code Review Bot [this message]
2026-04-12 4:32 ` Claude review: Introduce cold reset recovery method Claude Code Review Bot
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=review-patch4-20260406142325.157035-10-mallesh.koujalagi@intel.com \
--to=claude-review@example.com \
--cc=dri-devel-reviews@example.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox