From: Rodrigo Vivi <rodrigo.vivi@intel.com>
To: Stuart Summers <stuart.summers@intel.com>
Cc: <intel-xe@lists.freedesktop.org>, <matthew.brost@intel.com>,
<umesh.nerlige.ramappa@intel.com>, <gustavo.sousa@intel.com>,
<matthew.d.roper@intel.com>, <daniele.ceraolospurio@intel.com>,
<shuicheng.lin@intel.com>
Subject: Re: [PATCH 04/15] drm/xe: Invert vram_page_offline configfs attribute
Date: Tue, 13 Oct 2026 15:13:41 -0400 [thread overview]
Message-ID: <as6C5UM-OZsBx2YE@intel.com> (raw)
In-Reply-To: <20261005220636.602826-21-stuart.summers@intel.com>
On Mon, Oct 05, 2026 at 10:06:40PM +0000, Stuart Summers wrote:
> Rename disable_vram_page_offline to enable_vram_page_offline The default
> is now true, matching the previous default of offlining being active.
>
> Negative-logic flags are error prone once they get combined with other
> conditions, so carry the positive sense all the way down to the RAS
> consumer rather than inverting at the configfs boundary. The internal
> xe->ras field is explicitly initialized for every platform since a
> zeroed struct no longer means "offlining enabled".
yes, please!
I never forget our disable_power_well_enable where power_well_enable
already means power disabling! :)
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
>
> Signed-off-by: Stuart Summers <stuart.summers@intel.com>
> Assisted-by: LLM> ---
> drivers/gpu/drm/xe/xe_configfs.c | 44 +++++++++++++--------------
> drivers/gpu/drm/xe/xe_configfs.h | 2 +-
> drivers/gpu/drm/xe/xe_drm_ras_types.h | 4 +--
> drivers/gpu/drm/xe/xe_ras.c | 8 +++--
> drivers/gpu/drm/xe/xe_ttm_vram_mgr.c | 2 +-
> 5 files changed, 31 insertions(+), 29 deletions(-)
>
> diff --git a/drivers/gpu/drm/xe/xe_configfs.c b/drivers/gpu/drm/xe/xe_configfs.c
> index 6e62fdccb4c4..0ccf163306fc 100644
> --- a/drivers/gpu/drm/xe/xe_configfs.c
> +++ b/drivers/gpu/drm/xe/xe_configfs.c
> @@ -63,7 +63,7 @@
> * ├── engines_allowed
> * ├── migrate_ulls_period_ms
> * ├── enable_psmi
> - * └── disable_vram_page_offline
> + * └── enable_vram_page_offline
> *
> * After configuring the attributes as per next section, the device can be
> * probed with::
> @@ -161,15 +161,15 @@
> *
> * This attribute can only be set before binding to the device.
> *
> - * Disable VRAM page offline:
> - * ----------------------------
> + * Enable VRAM page offline:
> + * -------------------------
> *
> - * 0, n, N, false - Do not disable (Offlining is active - default)
> - * 1, y, Y, true - Disable vram page offline (Logging only)
> + * 0, n, N, false - Disable vram page offline (Logging only)
> + * 1, y, Y, true - Enable vram page offline (Offlining is active - default)
> *
> * Example to disable VRAM offline::
> *
> - * # echo 1 > /sys/kernel/config/xe/0000:03:00.0/disable_vram_page_offline
> + * # echo 0 > /sys/kernel/config/xe/0000:03:00.0/enable_vram_page_offline
> *
> * This attribute can only be set on CRI before binding to the device.
> *
> @@ -304,7 +304,7 @@ struct xe_config_group_device {
> bool survivability_mode;
> bool enable_psmi;
> bool enable_multi_queue;
> - bool disable_vram_page_offline;
> + bool enable_vram_page_offline;
> struct {
> unsigned int max_vfs;
> bool admin_only_pf;
> @@ -326,7 +326,7 @@ static const struct xe_config_device device_defaults = {
> .survivability_mode = false,
> .enable_psmi = false,
> .enable_multi_queue = true,
> - .disable_vram_page_offline = false,
> + .enable_vram_page_offline = true,
> .sriov = {
> .max_vfs = XE_DEFAULT_MAX_VFS,
> .admin_only_pf = XE_DEFAULT_ADMIN_ONLY_PF,
> @@ -675,15 +675,15 @@ static ssize_t migrate_ulls_period_ms_store(struct config_item *item,
> return len;
> }
>
> -static ssize_t disable_vram_page_offline_show(struct config_item *item, char *page)
> +static ssize_t enable_vram_page_offline_show(struct config_item *item, char *page)
> {
> struct xe_config_device *dev = to_xe_config_device(item);
>
> - return sprintf(page, "%s\n", str_yes_no(dev->disable_vram_page_offline));
> + return sprintf(page, "%s\n", str_yes_no(dev->enable_vram_page_offline));
> }
>
> -static ssize_t disable_vram_page_offline_store(struct config_item *item,
> - const char *page, size_t len)
> +static ssize_t enable_vram_page_offline_store(struct config_item *item,
> + const char *page, size_t len)
> {
> struct xe_config_group_device *dev = to_xe_config_group_device(item);
> bool val;
> @@ -697,7 +697,7 @@ static ssize_t disable_vram_page_offline_store(struct config_item *item,
> if (is_bound(dev))
> return -EBUSY;
>
> - dev->config.disable_vram_page_offline = val;
> + dev->config.enable_vram_page_offline = val;
>
> return len;
> }
> @@ -946,7 +946,7 @@ CONFIGFS_ATTR(, ctx_restore_post_bb);
> CONFIGFS_ATTR(, enable_multi_queue);
> CONFIGFS_ATTR(, migrate_ulls_period_ms);
> CONFIGFS_ATTR(, enable_psmi);
> -CONFIGFS_ATTR(, disable_vram_page_offline);
> +CONFIGFS_ATTR(, enable_vram_page_offline);
> CONFIGFS_ATTR(, engines_allowed);
> CONFIGFS_ATTR(, gt_types_allowed);
> CONFIGFS_ATTR(, survivability_mode);
> @@ -957,7 +957,7 @@ static struct configfs_attribute *xe_config_device_attrs[] = {
> &attr_enable_multi_queue,
> &attr_migrate_ulls_period_ms,
> &attr_enable_psmi,
> - &attr_disable_vram_page_offline,
> + &attr_enable_vram_page_offline,
> &attr_engines_allowed,
> &attr_gt_types_allowed,
> &attr_survivability_mode,
> @@ -989,7 +989,7 @@ static bool xe_config_device_is_visible(struct config_item *item,
> return false;
> }
>
> - if (attr == &attr_disable_vram_page_offline) {
> + if (attr == &attr_enable_vram_page_offline) {
> if (!dev->desc->is_dgfx || dev->desc->platform != XE_CRESCENTISLAND)
> return false;
> }
> @@ -1242,7 +1242,7 @@ static void dump_custom_dev_config(struct pci_dev *pdev,
> PRI_CUSTOM_ATTR("%u", migrate_ulls_period_ms);
> PRI_CUSTOM_ATTR("%d", enable_multi_queue);
> PRI_CUSTOM_ATTR("%d", enable_psmi);
> - PRI_CUSTOM_ATTR("%d", disable_vram_page_offline);
> + PRI_CUSTOM_ATTR("%d", enable_vram_page_offline);
> PRI_CUSTOM_ATTR("%d", survivability_mode);
> PRI_CUSTOM_ATTR("%u", sriov.admin_only_pf);
>
> @@ -1417,21 +1417,21 @@ u32 xe_configfs_get_migrate_ulls_period_ms(struct pci_dev *pdev)
> }
>
> /**
> - * xe_configfs_get_disable_vram_page_offline - get configfs disable_vram_page_offline setting
> + * xe_configfs_get_enable_vram_page_offline - get configfs enable_vram_page_offline setting
> * @pdev: pci device
> *
> - * Return: disable_vram_page_offline setting in configfs
> + * Return: enable_vram_page_offline setting in configfs
> */
> -bool xe_configfs_get_disable_vram_page_offline(struct pci_dev *pdev)
> +bool xe_configfs_get_enable_vram_page_offline(struct pci_dev *pdev)
> {
> struct xe_config_group_device *dev = find_xe_config_group_device(pdev);
> bool ret;
>
> if (!dev)
> - return device_defaults.disable_vram_page_offline;
> + return device_defaults.enable_vram_page_offline;
>
> scoped_guard(mutex, &dev->lock)
> - ret = dev->config.disable_vram_page_offline;
> + ret = dev->config.enable_vram_page_offline;
> config_group_put(&dev->group);
>
> return ret;
> diff --git a/drivers/gpu/drm/xe/xe_configfs.h b/drivers/gpu/drm/xe/xe_configfs.h
> index 14e49f23306f..1d538ce0ed2c 100644
> --- a/drivers/gpu/drm/xe/xe_configfs.h
> +++ b/drivers/gpu/drm/xe/xe_configfs.h
> @@ -25,7 +25,7 @@ u64 xe_configfs_get_engines_allowed(struct pci_dev *pdev);
> bool xe_configfs_get_psmi_enabled(struct pci_dev *pdev);
> bool xe_configfs_get_enable_multi_queue(struct pci_dev *pdev);
> u32 xe_configfs_get_migrate_ulls_period_ms(struct pci_dev *pdev);
> -bool xe_configfs_get_disable_vram_page_offline(struct pci_dev *pdev);
> +bool xe_configfs_get_enable_vram_page_offline(struct pci_dev *pdev);
> ssize_t xe_configfs_get_ctx_restore_mid_bb(struct pci_dev *pdev,
> enum xe_engine_class class,
> u32 *cs, size_t max_len);
> diff --git a/drivers/gpu/drm/xe/xe_drm_ras_types.h b/drivers/gpu/drm/xe/xe_drm_ras_types.h
> index 0be218ba2db7..60cbc35a6b35 100644
> --- a/drivers/gpu/drm/xe/xe_drm_ras_types.h
> +++ b/drivers/gpu/drm/xe/xe_drm_ras_types.h
> @@ -44,8 +44,8 @@ struct xe_drm_ras {
> /** @info: info array for all types of errors */
> struct xe_drm_ras_counter *info[DRM_XE_RAS_ERR_SEV_MAX];
>
> - /** @disable_vram_page_offline: cached configfs policy, immutable after init */
> - bool disable_vram_page_offline;
> + /** @enable_vram_page_offline: cached configfs policy, immutable after init */
> + bool enable_vram_page_offline;
> };
>
> #endif
> diff --git a/drivers/gpu/drm/xe/xe_ras.c b/drivers/gpu/drm/xe/xe_ras.c
> index 7a85735c57d5..ce9b57b3eede 100644
> --- a/drivers/gpu/drm/xe/xe_ras.c
> +++ b/drivers/gpu/drm/xe/xe_ras.c
> @@ -916,13 +916,15 @@ void xe_ras_init(struct xe_device *xe)
> {
> int ret;
>
> + xe->ras.enable_vram_page_offline = true;
> +
> /*
> - * TODO: Replace platform check with xe->info.has_disable_vram_page_offline
> + * TODO: Replace platform check with xe->info.has_vram_page_offline
> * once the feature flag is plumbed through device info.
> */
> if (xe->info.platform == XE_CRESCENTISLAND)
> - xe->ras.disable_vram_page_offline =
> - xe_configfs_get_disable_vram_page_offline(to_pci_dev(xe->drm.dev));
> + xe->ras.enable_vram_page_offline =
> + xe_configfs_get_enable_vram_page_offline(to_pci_dev(xe->drm.dev));
>
> xe_drm_ras_init(xe);
>
> diff --git a/drivers/gpu/drm/xe/xe_ttm_vram_mgr.c b/drivers/gpu/drm/xe/xe_ttm_vram_mgr.c
> index 9a514d983e90..1dad94a153b7 100644
> --- a/drivers/gpu/drm/xe/xe_ttm_vram_mgr.c
> +++ b/drivers/gpu/drm/xe/xe_ttm_vram_mgr.c
> @@ -933,7 +933,7 @@ int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr)
> vram_mgr = &vr->ttm;
> mm = &vram_mgr->mm;
>
> - if (xe->ras.disable_vram_page_offline) {
> + if (!xe->ras.enable_vram_page_offline) {
> xe_err(xe, "0x%llx is reported as corrupted address by HW\n",
> addr);
> return -EOPNOTSUPP;
> --
> 2.43.0
>
next prev parent reply other threads:[~2026-10-13 19:14 UTC|newest]
Thread overview: 28+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-10-05 22:06 [PATCH 00/15] Add new debug infrastructure for configfs Stuart Summers
2026-10-05 22:06 ` [PATCH 01/15] drm/xe: Guard configfs attribute reads in getters Stuart Summers
2026-10-05 22:06 ` [PATCH 02/15] drm/xe/configfs: Fix out-of-bounds read in parse_wa_bb_lines() Stuart Summers
2026-10-13 19:08 ` Rodrigo Vivi
2026-10-05 22:06 ` [PATCH 03/15] drm/xe/configfs: Copy wa_bb out under the configfs lock Stuart Summers
2026-10-13 19:10 ` Rodrigo Vivi
2026-10-13 20:35 ` Summers, Stuart
2026-10-05 22:06 ` [PATCH 04/15] drm/xe: Invert vram_page_offline configfs attribute Stuart Summers
2026-10-13 19:13 ` Rodrigo Vivi [this message]
2026-10-05 22:06 ` [PATCH 05/15] drm/xe: Make survivability_mode configfs attribute a bitmap Stuart Summers
2026-10-13 19:14 ` Rodrigo Vivi
2026-10-05 22:06 ` [PATCH 06/15] drm/xe: Sort xe_config_device fields Stuart Summers
2026-10-05 22:06 ` [PATCH 07/15] drm/xe: Split out configfs data structures Stuart Summers
2026-10-05 22:06 ` [PATCH 08/15] drm/xe: Add a new debug focused configfs group Stuart Summers
2026-10-05 22:06 ` [PATCH 09/15] drm/xe: Move debug configfs entries to xe_configfs_debug.c Stuart Summers
2026-10-13 19:16 ` Rodrigo Vivi
2026-10-05 22:06 ` [PATCH 10/15] drm/xe/guc: Add configfs support for guc_log_level Stuart Summers
2026-10-05 22:06 ` [PATCH 11/15] drm/xe/guc: Add support for NPK as a GuC log target Stuart Summers
2026-10-05 22:06 ` [PATCH 12/15] drm/xe: Add infrastructure for debug configfs parameters Stuart Summers
2026-10-05 22:06 ` [PATCH 13/15] drm/xe: Migrate existing debug configfs entries to params infrastructure Stuart Summers
2026-10-05 22:06 ` [PATCH 14/15] drm/xe: Taint kernel when debug configfs parameters are set Stuart Summers
2026-10-05 22:06 ` [PATCH 15/15] drm/xe: Add enable_media module parameter Stuart Summers
2026-10-05 22:13 ` ✗ CI.checkpatch: warning for Add new debug infrastructure for configfs (rev10) Patchwork
2026-10-05 22:15 ` ✓ CI.KUnit: success " Patchwork
2026-10-05 23:17 ` ✓ Xe.CI.BAT: " Patchwork
2026-10-06 6:22 ` ✗ Xe.CI.FULL: failure " Patchwork
2026-10-07 19:13 ` Summers, Stuart
-- strict thread matches above, loose matches on Subject: below --
2026-10-05 19:06 [PATCH 00/15] Add new debug infrastructure for configfs Stuart Summers
2026-10-05 19:06 ` [PATCH 04/15] drm/xe: Invert vram_page_offline configfs attribute Stuart Summers
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=as6C5UM-OZsBx2YE@intel.com \
--to=rodrigo.vivi@intel.com \
--cc=daniele.ceraolospurio@intel.com \
--cc=gustavo.sousa@intel.com \
--cc=intel-xe@lists.freedesktop.org \
--cc=matthew.brost@intel.com \
--cc=matthew.d.roper@intel.com \
--cc=shuicheng.lin@intel.com \
--cc=stuart.summers@intel.com \
--cc=umesh.nerlige.ramappa@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