Re: [RFC v5 5/5] drm/xe/RAS: send multicast event on occurrence of an error

[Aravind Iddamsetty <aravind.iddamsetty@linux.intel.com>]

[-- Attachment #1: Type: text/plain, Size: 2831 bytes --]


On 16-08-2025 03:31, Rodrigo Vivi wrote:
> On Wed, Jul 30, 2025 at 12:19:56PM +0530, Aravind Iddamsetty wrote:
>> Whenever a correctable or an uncorrectable error happens an event is sent
>> to the corresponding listeners of these groups.
>>
>> v2: Rebase
>> v3: protect with CONFIG_NET define.
>>
>> Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com> #v2
>> Signed-off-by: Aravind Iddamsetty <aravind.iddamsetty@linux.intel.com>
>> ---
>>  drivers/gpu/drm/xe/xe_hw_error.c | 41 ++++++++++++++++++++++++++++++++
>>  1 file changed, 41 insertions(+)
>>
>> diff --git a/drivers/gpu/drm/xe/xe_hw_error.c b/drivers/gpu/drm/xe/xe_hw_error.c
>> index bdd9c88674b2..e6e2e6250b70 100644
>> --- a/drivers/gpu/drm/xe/xe_hw_error.c
>> +++ b/drivers/gpu/drm/xe/xe_hw_error.c
>> @@ -2,6 +2,8 @@
>>  /*
>>   * Copyright © 2023 Intel Corporation
>>   */
>> +#include <net/genetlink.h>
>> +#include <uapi/drm/drm_netlink.h>
>>  
>>  #include "xe_gt_printk.h"
>>  #include "xe_hw_error.h"
>> @@ -776,6 +778,43 @@ xe_soc_hw_error_handler(struct xe_tile *tile, const enum hardware_error hw_err)
>>  				(HARDWARE_ERROR_MAX << 1) + 1);
>>  }
>>  
>> +#ifdef CONFIG_NET
>> +static void
>> +generate_netlink_event(struct xe_device *xe, const enum hardware_error hw_err)
>> +{
>> +	struct sk_buff *msg;
>> +	void *hdr;
>> +
>> +	if (!xe->drm.drm_genl_family)
>> +		return;
>> +
>> +	msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_ATOMIC);
>> +	if (!msg) {
>> +		drm_dbg_driver(&xe->drm, "couldn't allocate memory for error multicast event\n");
>> +		return;
>> +	}
>> +
>> +	hdr = genlmsg_put(msg, 0, 0, xe->drm.drm_genl_family, 0, DRM_RAS_CMD_ERROR_EVENT);
> this is something that could be wrapped up in the drm_ras
are you  referring to entire generate_netlink_event function? because I
thought driver might want to pass in custom info as part of event
may be like the error ID that was reported by HW.
Thanks,
Aravind.
>
>> +	if (!hdr) {
>> +		drm_dbg_driver(&xe->drm, "mutlicast msg buffer is small\n");
>> +		nlmsg_free(msg);
>> +		return;
>> +	}
>> +
>> +	genlmsg_end(msg, hdr);
>> +
>> +	genlmsg_multicast(xe->drm.drm_genl_family, msg, 0,
>> +			  hw_err ?
>> +			  DRM_GENL_MCAST_UNCORR_ERR
>> +			  : DRM_GENL_MCAST_CORR_ERR,
>> +			  GFP_ATOMIC);
>> +}
>> +#else
>> +static void
>> +generate_netlink_event(struct xe_device *xe, const enum hardware_error hw_err)
>> +{}
>> +#endif
>> +
>>  static void
>>  xe_hw_error_source_handler(struct xe_tile *tile, const enum hardware_error hw_err)
>>  {
>> @@ -837,6 +876,8 @@ xe_hw_error_source_handler(struct xe_tile *tile, const enum hardware_error hw_er
>>  	}
>>  
>>  	xe_mmio_write32(&tile->mmio, DEV_ERR_STAT_REG(hw_err), errsrc);
>> +
>> +	generate_netlink_event(tile_to_xe(tile), hw_err);
>>  unlock:
>>  	spin_unlock_irqrestore(&tile_to_xe(tile)->irq.lock, flags);
>>  }
>> -- 
>> 2.25.1
>>

[-- Attachment #2: Type: text/html, Size: 3793 bytes --]