Intel-XE Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: "Nilawar, Badal" <badal.nilawar@intel.com>
To: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: <intel-xe@lists.freedesktop.org>, <anshuman.gupta@intel.com>,
	<raag.jadav@intel.com>, <riana.tauro@intel.com>,
	<mallesh.koujalagi@intel.com>, <aravind.iddamsetty@intel.com>,
	<heikki.krogerus@linux.intel.com>,
	<himal.prasad.ghimiray@intel.com>
Subject: Re: [PATCH 1/2] drm/xe/i2c: Expose AMC Alert reason sysfs
Date: Wed, 9 Sep 2026 19:57:49 +0530	[thread overview]
Message-ID: <570c3776-1157-4d3f-985b-330caa7ca769@intel.com> (raw)
In-Reply-To: <aqBZXRgr6epDvWmE@intel.com>


On 09-09-2026 00:22, Rodrigo Vivi wrote:
> On Fri, Sep 04, 2026 at 09:36:10PM +0530, Badal Nilawar wrote:
>> AMC raises an SMBUS alert before performing a power removal or
>> power-cycle operation. The xe driver then places the device into
>> vendor-specific wedge mode until the recovery is performed.
>>
>> Expose a read-only xe_amc_alert_reason sysfs attribute to help users
>> identify the required recovery action.
>>
>> Assisted-by: Claude:claude-opus-4.8
>> Signed-off-by: Badal Nilawar <badal.nilawar@intel.com>
>> ---
>> v2:
>>   - Created sysfs during i2c probe time
>>   - Fix Documentation (Rodrigo)
>> ---
>>   .../ABI/testing/sysfs-driver-intel-xe-amc     | 21 +++++++
>>   drivers/gpu/drm/xe/xe_amc.c                   | 61 ++++++++++++++++++-
>>   2 files changed, 79 insertions(+), 3 deletions(-)
>>   create mode 100644 Documentation/ABI/testing/sysfs-driver-intel-xe-amc
>>
>> diff --git a/Documentation/ABI/testing/sysfs-driver-intel-xe-amc b/Documentation/ABI/testing/sysfs-driver-intel-xe-amc
>> new file mode 100644
>> index 000000000000..4e5389163dbb
>> --- /dev/null
>> +++ b/Documentation/ABI/testing/sysfs-driver-intel-xe-amc
>> @@ -0,0 +1,21 @@
>> +What:		/sys/bus/pci/drivers/xe/.../xe_amc_alert_reason
>> +Date:		September 2026
>> +KernelVersion:	7.4
>> +Contact:	intel-xe@lists.freedesktop.org
>> +Description:
>> +		This file exposes the reason for the most recent Add-In
>> +		Management Controller (AMC) alert on Intel Xe platforms.
>> +
>> +		An AMC alert is delivered via an SMBUS interrupt and causes the
>> +		device to be wedged, requiring vendor-specific recovery. This
>> +		attribute is created when such an alert is handled and is
>> +		available to all users as read-only.
>> +
>> +		Read returns a single line containing one of the following
>> +		alert reasons:
>> +
>> +			Firmware Download
>> +			Thermal Trip
>> +			OOB Request
>> +			OOB Reset
>> +			Catastrophic
> please make this non capital letters and no space
>
> firmware_download
> thermal_trip
> oob_request
> oob_reset
> catastrophic
>
>> diff --git a/drivers/gpu/drm/xe/xe_amc.c b/drivers/gpu/drm/xe/xe_amc.c
>> index 8ecadee6eea3..c2d516c76502 100644
>> --- a/drivers/gpu/drm/xe/xe_amc.c
>> +++ b/drivers/gpu/drm/xe/xe_amc.c
>> @@ -18,6 +18,7 @@
>>   #include "xe_device.h"
>>   #include "xe_i2c.h"
>>   #include "xe_mmio.h"
>> +#include "xe_printk.h"
>>   
>>   /**
>>    * DOC: Add-In Management Controller (AMC)
>> @@ -46,6 +47,7 @@ enum xe_amc_alert {
>>   };
>>   
>>   static const char * const amc_alert[] = {
>> +	[AMC_ALERT_UNKNOWN]		= "None",
> shouldn't be unknown?
It can be unknown.
> shouldn't be in the list above?

Not required, as only valid alert reasons result in the device being 
wedged and subsequently exposed to userspace.

Thanks,
Badal

>
>>   	[AMC_ALERT_FW_DOWNLOAD]		= "Firmware Download",
>>   	[AMC_ALERT_THERMAL_TRIP]	= "Thermal Trip",
>>   	[AMC_ALERT_OOB_REQUEST]		= "OOB Request",
>> @@ -56,6 +58,8 @@ static const char * const amc_alert[] = {
>>   struct xe_amc {
>>   	struct xe_i2c *i2c;
>>   	struct work_struct work;
>> +	u8 alert_reason;
>> +	bool sysfs_created;
>>   };
>>   
>>   struct amc_header {
>> @@ -104,6 +108,45 @@ static const struct amc_request amc_get_alert_reason = {
>>   	},
>>   };
>>   
>> +/**
>> + * DOC: AMC Alert Reason
>> + *
>> + * On Intel Xe platforms, AMC sends an alert notification via an SMBUS interrupt
>> + * to notify events such as firmware download, thermal trip or a
>> + * catastrophic error. See enum xe_amc_alert for the full list of reasons.
>> + * Upon an AMC alert the device is wedged and requires vendor-specific recovery.
>> + *
>> + * The alert reason is exposed through
>> + * /sys/bus/pci/devices/<BDF>/xe_amc_alert_reason
>> + *
>> + * See Documentation/ABI/testing/sysfs-driver-intel-xe-amc for the ABI
>> + * specification.
>> + */
>> +
>> +static ssize_t xe_amc_alert_reason_show(struct device *dev,
>> +					struct device_attribute *attr, char *buff)
>> +{
>> +	struct xe_device *xe = pdev_to_xe_device(to_pci_dev(dev));
>> +	struct xe_amc *amc = xe->i2c->amc;
>> +
>> +	return sysfs_emit(buff, "%s\n", amc_alert[amc->alert_reason]);
>> +}
>> +static DEVICE_ATTR_RO(xe_amc_alert_reason);
>> +
>> +static void xe_amc_remove_alert_sysfs(struct xe_i2c *i2c)
>> +{
>> +	struct device *dev = i2c->drm_dev;
>> +
>> +	device_remove_file(dev, &dev_attr_xe_amc_alert_reason);
>> +}
>> +
>> +static int xe_amc_create_alert_sysfs(struct xe_i2c *i2c)
>> +{
>> +	struct device *dev = i2c->drm_dev;
>> +
>> +	return device_create_file(dev, &dev_attr_xe_amc_alert_reason);
>> +}
>> +
>>   static void xe_amc_work(struct work_struct *work)
>>   {
>>   	const struct amc_request *request = &amc_get_alert_reason;
>> @@ -158,10 +201,15 @@ static void xe_amc_work(struct work_struct *work)
>>   	case AMC_ALERT_THERMAL_TRIP:
>>   	case AMC_ALERT_OOB_REQUEST:
>>   	case AMC_ALERT_OOB_RESET:
>> -	case AMC_ALERT_CATERR:
>> +	case AMC_ALERT_CATERR: {
>> +		struct xe_device *xe = i2c_client_to_xe_device(client);
>> +
>>   		dev_warn(amc->i2c->drm_dev, "AMC Alert: %s\n", amc_alert[alert_reason]);
>> -		xe_device_declare_wedged(i2c_client_to_xe_device(client));
>> +		amc->alert_reason = alert_reason;
>> +		xe_device_set_wedged_method(xe, DRM_WEDGE_RECOVERY_VENDOR);
>> +		xe_device_declare_wedged(xe);
>>   		break;
>> +	}
>>   	default:
>>   		dev_warn(amc->i2c->drm_dev, "unknown AMC alert: %d\n", alert_reason);
>>   		break;
>> @@ -176,6 +224,7 @@ void xe_amc_handle_alert(struct xe_i2c *i2c)
>>   int xe_amc_init(struct xe_i2c *i2c)
>>   {
>>   	struct xe_amc *amc;
>> +	int ret;
>>   
>>   	amc = kzalloc(sizeof(*amc), GFP_KERNEL);
>>   	if (!amc)
>> @@ -185,11 +234,17 @@ int xe_amc_init(struct xe_i2c *i2c)
>>   	i2c->amc = amc;
>>   	amc->i2c = i2c;
>>   
>> -	return 0;
>> +	ret =  xe_amc_create_alert_sysfs(i2c);
>> +	if (ret)
>> +		kfree(i2c->amc);
>> +
>> +	return ret;
>>   }
>>   
>>   void xe_amc_exit(struct xe_i2c *i2c)
>>   {
>> +	xe_amc_remove_alert_sysfs(i2c);
>> +
>>   	if (i2c->amc) {
>>   		cancel_work_sync(&i2c->amc->work);
>>   		kfree(i2c->amc);
>> -- 
>> 2.54.0
>>

  reply	other threads:[~2026-09-09 14:28 UTC|newest]

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-09-04 16:06 [PATCH 0/2] Expose device sysfs for AMC and GPU UUID Badal Nilawar
2026-09-04 15:56 ` ✗ CI.checkpatch: warning for " Patchwork
2026-09-04 15:58 ` ✓ CI.KUnit: success " Patchwork
2026-09-04 16:06 ` [PATCH 1/2] drm/xe/i2c: Expose AMC Alert reason sysfs Badal Nilawar
2026-09-08 18:52   ` Rodrigo Vivi
2026-09-09 14:27     ` Nilawar, Badal [this message]
2026-09-09 15:32       ` Nilawar, Badal
2026-09-04 16:06 ` [PATCH 2/2] drm/xe/cri: Expose device UUID through sysfs Badal Nilawar
2026-09-04 15:57   ` sashiko-bot
2026-09-04 16:02   ` Gupta, Anshuman
2026-09-04 16:37   ` Michal Wajdeczko
2026-09-07  9:13     ` Nilawar, Badal
2026-09-07 11:55       ` Michal Wajdeczko
2026-09-07 12:04         ` Gupta, Anshuman
2026-09-07 15:48           ` Michal Wajdeczko
2026-09-08 18:55             ` Rodrigo Vivi
2026-09-10  6:31         ` Nilawar, Badal
2026-09-07  5:02   ` Joonas Lahtinen
2026-09-07  8:16     ` Nilawar, Badal
2026-09-04 16:48 ` ✓ Xe.CI.BAT: success for Expose device sysfs for AMC and GPU UUID Patchwork
2026-09-05  0:54 ` ✓ Xe.CI.FULL: " Patchwork

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=570c3776-1157-4d3f-985b-330caa7ca769@intel.com \
    --to=badal.nilawar@intel.com \
    --cc=anshuman.gupta@intel.com \
    --cc=aravind.iddamsetty@intel.com \
    --cc=heikki.krogerus@linux.intel.com \
    --cc=himal.prasad.ghimiray@intel.com \
    --cc=intel-xe@lists.freedesktop.org \
    --cc=mallesh.koujalagi@intel.com \
    --cc=raag.jadav@intel.com \
    --cc=riana.tauro@intel.com \
    --cc=rodrigo.vivi@intel.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