Re: [PATCH v4 11/32] drm/xe/log: Index all SIGID printk messages

[Michal Wajdeczko <michal.wajdeczko@intel.com>]

On 8/13/2026 2:54 PM, Michal Wajdeczko wrote:
> 
> 
> On 8/13/2026 2:31 PM, Mallesh, Koujalagi wrote:
>>
>> On 13-08-2026 12:44 am, Michal Wajdeczko wrote:
>>> When CONFIG_PRINTK_INDEX is enabled, it is expected that all device
>>> level printk messages are indexed for audit. While usually this is
>>> done automatically behind the scenes when code is using regular
>>> dev_printk macros, since we are generating different dmesg messages
>>> inside xe_log_emit() based on the severity, component and location,
>>> we only get those entries in /sys/kernel/debug/printk/index/xe:
>>>
>>>    <3> drivers/gpu/drm/xe/xe_log.c:142 log_dmesg_vprintk "%s %s: [drm] *ERROR* %pV"
>>>    <6> drivers/gpu/drm/xe/xe_log.c:140 log_dmesg_vprintk "%s %s: [drm] %pV"
>>>
>>> Explicitly generate printk index using dev_printk_index_emit() with
>>> some generic prefix that includes the SIGID tag.
>>>
>>> Suggested-by: Jani Nikula <jani.nikula@intel.com>
>>> Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
>>> Cc: Jani Nikula <jani.nikula@intel.com>
>>> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
>>> ---
>>>   drivers/gpu/drm/xe/xe_log.c |  8 ++++----
>>>   drivers/gpu/drm/xe/xe_log.h | 14 +++++++++++---
>>>   2 files changed, 15 insertions(+), 7 deletions(-)
>>>
>>> diff --git a/drivers/gpu/drm/xe/xe_log.c b/drivers/gpu/drm/xe/xe_log.c
>>> index 0b423ed121cd..50a9b35e5b3e 100644
>>> --- a/drivers/gpu/drm/xe/xe_log.c
>>> +++ b/drivers/gpu/drm/xe/xe_log.c
>>> @@ -177,7 +177,7 @@ static void log_emit_dmesg(struct pci_dev *pdev, int cper_sev, enum xe_sigid sig
>>>   }
>>>     /**
>>> - * xe_log_emit() - Emit a structured SIGID log entry
>>> + * __xe_log_emit() - Emit a structured SIGID log entry
>>>    * @pdev: the &pci_dev device
>>>    * @cper_sev: CPER severity (CPER_SEV_FATAL, CPER_SEV_RECOVERABLE, ...)
>>>    * @sigid: signature identifier, see &enum xe_sigid
>>> @@ -206,9 +206,9 @@ static void log_emit_dmesg(struct pci_dev *pdev, int cper_sev, enum xe_sigid sig
>>>    *   <3> xe 0000:03:00.0: [drm] *ERROR* SIGID=106 (-ETIMEDOUT) Engine 'rcs0' hung
>>>    *   <6> xe 0000:03:00.0: [drm] SIGID=103 In survivability mode
>>>    */
>>> -void xe_log_emit(struct pci_dev *pdev, int cper_sev, enum xe_sigid sigid,
>>> -         u32 component, u32 location, const void *data, size_t len,
>>> -         const char *fmt, ...)
>>> +void __xe_log_emit(struct pci_dev *pdev, int cper_sev, enum xe_sigid sigid,
>>> +           u32 component, u32 location, const void *data, size_t len,
>>> +           const char *fmt, ...)
>>>   {
>>>       struct va_format vaf;
>>>       va_list args;
>>> diff --git a/drivers/gpu/drm/xe/xe_log.h b/drivers/gpu/drm/xe/xe_log.h
>>> index 0928b0866617..53fe2bb7ddd3 100644
>>> --- a/drivers/gpu/drm/xe/xe_log.h
>>> +++ b/drivers/gpu/drm/xe/xe_log.h
>>> @@ -16,9 +16,17 @@
>>>   struct pci_dev;
>>>     __printf(8, 9)
>>> -void xe_log_emit(struct pci_dev *pdev, int cper_sev, enum xe_sigid sigid,
>>> -         u32 component, u32 location, const void *data, size_t len,
>>> -         const char *fmt, ...);
>>> +void __xe_log_emit(struct pci_dev *pdev, int cper_sev, enum xe_sigid sigid,
>>> +           u32 component, u32 location, const void *data, size_t len,
>>> +           const char *fmt, ...);
>>> +
>>> +#define __xe_log_emit_printk_index(fmt) \
>>> +    dev_printk_index_emit(NULL, "[drm]%s SIGID=%u %s(%s)%s%s%s: " fmt);
>> We need to change format for a. error path b. blob path c. plain or info path right?
> 
> well, that's not doable as final format/output depends on the
> severity and data value, which could be non-const at compile
> time, while printk-index requires this to emit right entry
> 
> I can change that to a something more generic (and unfriendly), like:
> 
>     dev_printk_index_emit(NULL, "%sSIGID=%u %s" fmt);
> 
> but that will catch and match all our outputs.

and this is how this will look like for current series:

$ sudo cat /sys/kernel/debug/printk/index/xe | grep SIGID
<3> drivers/gpu/drm/xe/xe_device.c:1481 xe_device_declare_wedged "%s %s: %s SIGID=%u %sDevice declared wedged!\n"
<3> drivers/gpu/drm/xe/xe_gt.c:977 gt_reset_worker "%s %s: %s SIGID=%u %sreset failed\n"
<6> drivers/gpu/drm/xe/xe_gt.c:968 gt_reset_worker "%s %s: %s SIGID=%u %sreset done\n"
<6> drivers/gpu/drm/xe/xe_gt.c:929 gt_reset_worker "%s %s: %s SIGID=%u %sreset started\n"
<3> drivers/gpu/drm/xe/xe_guc.c:1620 xe_guc_mmio_send_recv "%s %s: %s SIGID=%u %sMMIO request %#x: unexpected reply %#x\n"
<3> drivers/gpu/drm/xe/xe_guc.c:1612 xe_guc_mmio_send_recv "%s %s: %s SIGID=%u %sMMIO request %#x: failure %#x hint %#x\n"
<3> drivers/gpu/drm/xe/xe_guc.c:1556 xe_guc_mmio_send_recv "%s %s: %s SIGID=%u %sMMIO request %#x: no reply %#x\n"
<3> drivers/gpu/drm/xe/xe_guc.c:1546 xe_guc_mmio_send_recv "%s %s: %s SIGID=%u %sMMIO request %#x: lost, too many retries %u\n"
<3> drivers/gpu/drm/xe/xe_pagefault.c:631 xe_pagefault_queue_work "%s %s: %s SIGID=%u %sUnsuccessful response\n"
<3> drivers/gpu/drm/xe/xe_pagefault.c:332 xe_pagefault_queue_add "%s %s: %s SIGID=%u %sQueue full!\n"
<3> drivers/gpu/drm/xe/xe_pci.c:1181 xe_pci_probe "%s %s: %s SIGID=%u %sdriver loading failed for device '%04x'\n"
<6> drivers/gpu/drm/xe/xe_pci.c:1171 xe_pci_probe "%s %s: %s SIGID=%u %sdriver loading blocked for device '%04x'\n"
<3> drivers/gpu/drm/xe/xe_pci_error.c:102 xe_pci_error_slot_reset "%s %s: %s SIGID=%u %sCannot re-enable PCI device after reset\n"
<3> drivers/gpu/drm/xe/xe_pcode.c:321 xe_pcode_ready "%s %s: %s SIGID=%u %sinitialization timedout after %ld seconds\n"
<3> drivers/gpu/drm/xe/xe_pcode.c:221 xe_pcode_request "%s %s: %s SIGID=%u %stimeout, retrying with preemption disabled\n"
<3> drivers/gpu/drm/xe/xe_pcode.c:65 pcode_mailbox_status "%s %s: %s SIGID=%u %sMailbox failed: %s\n"
<6> drivers/gpu/drm/xe/xe_survivability_mode.c:440 log_survivability_info "%s %s: %s SIGID=%u %s%s: %#x\n"
<3> drivers/gpu/drm/xe/xe_survivability_mode.c:433 log_survivability_info "%s %s: %s SIGID=%u %sBoot Status: %s (%u)\n"
<6> drivers/gpu/drm/xe/xe_survivability_mode.c:422 xe_survivability_mode_runtime_enable "%s %s: %s SIGID=%u %sFirmware flash required!\n"
<6> drivers/gpu/drm/xe/xe_survivability_mode.c:417 xe_survivability_mode_runtime_enable "%s %s: %s SIGID=%u %sRuntime Mode enabled!\n"
<3> drivers/gpu/drm/xe/xe_survivability_mode.c:409 xe_survivability_mode_runtime_enable "%s %s: %s SIGID=%u %sRuntime Mode not supported!\n"
<3> drivers/gpu/drm/xe/xe_survivability_mode.c:344 enable_boot_survivability_mode "%s %s: %s SIGID=%u %sFailed to enable Boot Mode!\n"
<6> drivers/gpu/drm/xe/xe_survivability_mode.c:339 enable_boot_survivability_mode "%s %s: %s SIGID=%u %sBoot Mode enabled!\n"
<3> drivers/gpu/drm/xe/xe_survivability_mode.c:334 enable_boot_survivability_mode "%s %s: %s SIGID=%u %sBoot Mode enabled!\n"
<3> drivers/gpu/drm/xe/xe_survivability_mode.c:302 create_survivability_sysfs "%s %s: %s SIGID=%u %sFailed to create sysfs files!\n"


> 
> @Jani, are you OK with that?
> 
>>> +
>>> +#define xe_log_emit(pdev, sev, sig, comp, loc, data, len, fmt, args...) ({        \
>>> +    __xe_log_emit_printk_index(fmt);                        \
>>> +    __xe_log_emit((pdev), (sev), (sig), (comp), (loc), (data), (len), fmt, ##args); \
>>> +})
>>
>> nit: use do{} while (0)
>>
>> Reviewed-by: Mallesh Koujalagi <mallesh.koujalagi@intel.com>
>>
>>>     #define xe_log_emit_fatal(pdev, sig, comp, loc, data, len, fmt, args...) \
>>>       xe_log_emit((pdev), CPER_SEV_FATAL, (sig), (comp), (loc), \
>