All of lore.kernel.org
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Daniele Ceraolo Spurio" <daniele.ceraolospurio@intel.com>,
	intel-xe@lists.freedesktop.org
Subject: Re: [PATCH] drm/xe: Add GuC FW path configfs overrides
Date: Thu, 19 Nov 2026 02:00:47 +0000	[thread overview]
Message-ID: <sashiko-outbox-222315@kernel.org> (raw)
In-Reply-To: <20261119014602.243548-2-daniele.ceraolospurio@intel.com>

Thank you for your contribution! Sashiko AI review found 4 potential
issues to consider:
- [High] drm/xe/configfs: 1-byte OOB write in __gt_guc_override_show()
- [Medium] drm/xe/configfs: empty override in __gt_guc_override_store()
  disables GuC
- [Medium] drm/xe: unchecked drmm_kstrdup() failure in
  xe_configfs_guc_override()
- [Low] drm/xe/configfs: missing guc_overrides in
  dump_custom_dev_config()
--

commit f425c98fc376b284cd705ef5abb819f7ae2fc52d
Author: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>

drm/xe: Add GuC FW path configfs overrides

This commit adds per-GT configfs attributes (primary_gt_guc_override and
media_gt_guc_override) to override the GuC firmware path independently for
the primary and media GTs.

> diff --git a/drivers/gpu/drm/xe/xe_configfs.c b/drivers/gpu/drm/xe/xe_configfs.c
> index 2ddd707a6dbd3..b6b544c3bc9da 100644
> --- a/drivers/gpu/drm/xe/xe_configfs.c
> +++ b/drivers/gpu/drm/xe/xe_configfs.c
[ ... ]
> @@ -309,6 +329,11 @@ struct xe_config_group_device {
>  			unsigned int max_vfs;
>  			bool admin_only_pf;
>  		} sriov;
> +
> +		struct {
> +			char *primary_gt;
> +			char *media_gt;
> +		} guc_overrides;
>  	} config;
>  
>  	/* protects attributes */
> @@ -331,6 +356,10 @@ static const struct xe_config_device device_defaults = {
>  		.max_vfs = XE_DEFAULT_MAX_VFS,
>  		.admin_only_pf = XE_DEFAULT_ADMIN_ONLY_PF,
>  	},
> +	.guc_overrides = {
> +		.primary_gt = NULL,
> +		.media_gt = NULL,
> +	},
>  };

[Severity: Low]
Should dump_custom_dev_config() also print guc_overrides.primary_gt and
guc_overrides.media_gt?  When either override is set, memcmp() in
xe_configfs_check_device() logs "Found custom settings in configfs" and
calls dump_custom_dev_config(), which does not include the new fields.

> @@ -937,6 +966,84 @@ static ssize_t ctx_restore_post_bb_store(struct config_item *item,
>  	return wa_bb_store(dev->config.ctx_restore_post_bb, dev, data, sz);
>  }
>  
> +static ssize_t __gt_guc_override_show(struct xe_config_group_device *dev,
> +				      char **override, char *page)
> +{
> +	guard(mutex)(&dev->lock);
> +
> +	return sprintf(page, "%s\n", *override ? *override : "no override");
> +}
> +
> +static ssize_t __gt_guc_override_store(struct xe_config_group_device *dev,
> +				       char **override, const char *page, size_t len)
> +{
> +	char *path;
> +
> +	if (len > PATH_MAX)
> +		return -EINVAL;

[Severity: High]
Can __gt_guc_override_show() write 1 byte past the end of the PAGE_SIZE
configfs read buffer?

In fs/configfs/file.c, fill_write_buffer() caps writes at
SIMPLE_ATTR_SIZE - 1 (4095) bytes, so len > PATH_MAX (4096) in
__gt_guc_override_store() accepts a 4095-byte string without a trailing
newline and stores a 4095-character string in *override.  When the
attribute is read back, fill_read_buffer() passes a PAGE_SIZE (4096-byte)
buffer to __gt_guc_override_show(), and sprintf(page, "%s\n", *override)
writes 4095 characters, '\n', and '\0' (4097 bytes).  Using sysfs_emit()
in __gt_guc_override_show() and rejecting len >= PATH_MAX in
__gt_guc_override_store() would avoid this overflow.

> +
> +	guard(mutex)(&dev->lock);
> +	if (is_bound(dev))
> +		return -EBUSY;
> +
> +	/*
> +	 * An empty string disables the override. Some bash commands can add a
> +	 * '\n' or '\0 ' at the end of the string, so consider a string of
> +	 * length 1 containing one of those characters as empty.
> +	 */
> +	if (!len || (len == 1 && (page[0] == '\n' || page[0] == '\0'))) {
> +		kfree(*override);
> +		*override = NULL;
> +	} else {
> +		path = krealloc(*override, len + 1, GFP_KERNEL);
> +		if (!path)
> +			return -ENOMEM;
> +
> +		memcpy(path, page, len);
> +
> +		/*
> +		 * the echo command adds a trailing newline unless the -n option
> +		 * is used, so replace it if present.
> +		 */
> +		if (path[len - 1] == '\n')
> +			path[len - 1] = '\0';
> +		else
> +			path[len] = '\0';
> +
> +		*override = path;
> +	}

[Severity: Medium]
Can writing a NUL byte followed by a newline (e.g. "echo -e '\0'", where
len == 2 and page[0] == '\0') store a non-NULL empty string in *override?

Because the empty-string check requires len == 1, a 2-byte "\0\n" input
takes the else branch and stores a buffer starting with '\0'.  During
probe, xe_configfs_guc_override() duplicates that empty string and
uc_fw_override() sets uc_fw->path to "", which causes uc_fw_request() to
mark GuC firmware as XE_UC_FIRMWARE_DISABLED and xe_wopcm_init() to fail
probe with -EINVAL.  Checking page[0] == '\0' regardless of len (or
checking !strlen(path)) would treat any NUL-leading write as empty.

[ ... ]

> @@ -1479,6 +1592,43 @@ u32 xe_configfs_get_ctx_restore_post_bb(struct pci_dev *pdev,
>  	return len;
>  }
>  
> +/**
> + * xe_configfs_guc_override - get configfs guc override path
> + * @pdev: pci device
> + * @gt_type: type of the GT for which to get the guc override path
> + *
> + * Return: A new drmm-managed allocation containing the path to the specified GT
> + * guc override if set, NULL otherwise
> + */
> +const char *xe_configfs_guc_override(struct pci_dev *pdev, enum xe_gt_type gt_type)
> +{
> +	struct xe_config_group_device *dev = find_xe_config_group_device(pdev);
> +	struct drm_device *drm = pci_get_drvdata(pdev);
> +	const char *override, *path = NULL;
> +
> +	if (!dev)
> +		return NULL;
> +
> +	if (!drm)
> +		goto out_put;
> +
> +	scoped_guard(mutex, &dev->lock) {
> +		if (gt_type == XE_GT_TYPE_MAIN)
> +			override = dev->config.guc_overrides.primary_gt;
> +		else if (gt_type == XE_GT_TYPE_MEDIA)
> +			override = dev->config.guc_overrides.media_gt;
> +		else
> +			override = NULL;
> +
> +		if (override)
> +			path = drmm_kstrdup(drm, override, GFP_KERNEL);
> +	}
> +
> +out_put:
> +	config_group_put(&dev->group);
> +	return path;
> +}

[Severity: Medium]
If drmm_kstrdup() fails, does xe_configfs_guc_override() return NULL and
make the allocation failure indistinguishable from having no override
configured?  uc_fw_override() will then leave uc_fw->path set to the
default auto-selected firmware path and silently load the default firmware
instead of failing probe with -ENOMEM.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261119014602.243548-2-daniele.ceraolospurio@intel.com?part=1

  reply	other threads:[~2026-11-19  2:01 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-19  1:46 [PATCH] drm/xe: Add GuC FW path configfs overrides Daniele Ceraolo Spurio
2026-11-19  2:00 ` sashiko-bot [this message]
2026-11-19  6:49 ` ✓ CI.KUnit: success for " Patchwork
2026-11-19  7:25 ` ✓ Xe.CI.BAT: " Patchwork
2026-11-20 11:00 ` ✗ Xe.CI.FULL: failure " Patchwork
2026-11-20 15:28 ` [PATCH] " Gustavo Sousa
2026-11-20 18:58   ` Ceraolo Spurio, Daniele

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=sashiko-outbox-222315@kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=daniele.ceraolospurio@intel.com \
    --cc=intel-xe@lists.freedesktop.org \
    --cc=sashiko-reviews@lists.linux.dev \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.