From: Lucas De Marchi <lucas.demarchi@intel.com>
To: intel-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org
Cc: "Thomas Hellström" <thomas.hellstrom@linux.intel.com>,
"Lucas De Marchi" <lucas.demarchi@intel.com>,
"Sumit Semwal" <sumit.semwal@linaro.org>,
"Thomas Zimmermann" <tzimmermann@suse.de>,
"Christian König" <christian.koenig@amd.com>
Subject: [Intel-gfx] [PATCH v3 00/16] drm/i915/guc: Refactor ADS access to use iosys_map
Date: Wed, 16 Feb 2022 09:41:31 -0800 [thread overview]
Message-ID: <20220216174147.3073235-1-lucas.demarchi@intel.com> (raw)
original: https://patchwork.freedesktop.org/series/99378/
v2: https://patchwork.freedesktop.org/series/99711/#rev1,
https://patchwork.freedesktop.org/series/99711/#rev2
Main changes from previous version:
- Unrelated patches to iosys-map conversion have landed
- Remove unecessary kernel.h include from iosys-map.h
- Rebase on latest drm-tip
Original cover letter:
While porting i915 to arm64 we noticed some issues accessing lmem.
Some writes were getting corrupted and the final state of the buffer
didn't have exactly what we wrote. This became evident when enabling
GuC submission: depending on the number of engines the ADS struct was
being corrupted and GuC would reject it, refusin to initialize.
From Documentation/core-api/bus-virt-phys-mapping.rst:
This memory is called "PCI memory" or "shared memory" or "IO memory" or
whatever, and there is only one way to access it: the readb/writeb and
related functions. You should never take the address of such memory, because
there is really nothing you can do with such an address: it's not
conceptually in the same memory space as "real memory" at all, so you cannot
just dereference a pointer. (Sadly, on x86 it **is** in the same memory space,
so on x86 it actually works to just deference a pointer, but it's not
portable).
When reading or writing words directly to IO memory, in order to be portable
the Linux kernel provides the abstraction detailed in section "Differences
between I/O access functions" of Documentation/driver-api/device-io.rst.
This limits our ability to simply overlay our structs on top a buffer
and directly access it since that buffer may come from IO memory rather than
system memory. Hence the approach taken in intel_guc_ads.c needs to be
refactored. This is not the only place in i915 that need to be changed, but
the one causing the most problems, with a real reproducer. This first set of
patch focuses on fixing the gem object to pass the ADS
After the addition of a few helpers in the dma_buf_map API, most of
intel_guc_ads.c can be converted to use it. The exception is the regset
initialization: we'd incur into a lot of extra indirection when
reading/writing each register. So the regset is converted to use a
temporary buffer allocated on probe, which is then copied to its
final location when finishing the initialization or on gt reset.
[v3: the part unrelated to iosys-map has already landed]
Testing on some discrete cards, after this change we can correctly pass the
ADS struct to GuC and have it initialized correctly.
thanks
Lucas De Marchi
Lucas De Marchi (16):
iosys-map: Add offset to iosys_map_memcpy_to()
iosys-map: Add a few more helpers
drm/i915/gt: Add helper for shmem copy to iosys_map
drm/i915/guc: Keep iosys_map of ads_blob around
drm/i915/guc: Add read/write helpers for ADS blob
drm/i915/guc: Convert golden context init to iosys_map
drm/i915/guc: Convert policies update to iosys_map
drm/i915/guc: Convert engine record to iosys_map
drm/i915/guc: Convert guc_ads_private_data_reset to iosys_map
drm/i915/guc: Convert golden context prep to iosys_map
drm/i915/guc: Replace check for golden context size
drm/i915/guc: Convert mapping table to iosys_map
drm/i915/guc: Convert capture list to iosys_map
drm/i915/guc: Convert guc_mmio_reg_state_init to iosys_map
drm/i915/guc: Convert __guc_ads_init to iosys_map
drm/i915/guc: Remove plain ads_blob pointer
drivers/gpu/drm/drm_cache.c | 2 +-
drivers/gpu/drm/drm_fb_helper.c | 2 +-
drivers/gpu/drm/i915/gt/shmem_utils.c | 32 +++
drivers/gpu/drm/i915/gt/shmem_utils.h | 3 +
drivers/gpu/drm/i915/gt/uc/intel_guc.h | 7 +-
drivers/gpu/drm/i915/gt/uc/intel_guc_ads.c | 233 ++++++++++--------
drivers/gpu/drm/i915/gt/uc/intel_guc_ads.h | 3 +-
.../gpu/drm/i915/gt/uc/intel_guc_submission.c | 17 +-
include/linux/iosys-map.h | 218 +++++++++++++++-
9 files changed, 396 insertions(+), 121 deletions(-)
--
2.35.1
next reply other threads:[~2022-02-16 17:40 UTC|newest]
Thread overview: 37+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-02-16 17:41 Lucas De Marchi [this message]
2022-02-16 17:41 ` [Intel-gfx] [PATCH v3 01/16] iosys-map: Add offset to iosys_map_memcpy_to() Lucas De Marchi
2022-02-16 17:41 ` [Intel-gfx] [PATCH v3 02/16] iosys-map: Add a few more helpers Lucas De Marchi
2022-02-17 8:42 ` Thomas Zimmermann
2022-02-20 4:28 ` Lucas De Marchi
2022-02-16 17:41 ` [Intel-gfx] [PATCH v3 03/16] drm/i915/gt: Add helper for shmem copy to iosys_map Lucas De Marchi
2022-02-16 17:41 ` [Intel-gfx] [PATCH v3 04/16] drm/i915/guc: Keep iosys_map of ads_blob around Lucas De Marchi
2022-02-16 17:41 ` [Intel-gfx] [PATCH v3 05/16] drm/i915/guc: Add read/write helpers for ADS blob Lucas De Marchi
2022-02-16 17:41 ` [Intel-gfx] [PATCH v3 06/16] drm/i915/guc: Convert golden context init to iosys_map Lucas De Marchi
2022-02-16 17:41 ` [Intel-gfx] [PATCH v3 07/16] drm/i915/guc: Convert policies update " Lucas De Marchi
2022-02-16 17:41 ` [Intel-gfx] [PATCH v3 08/16] drm/i915/guc: Convert engine record " Lucas De Marchi
2022-02-16 17:41 ` [Intel-gfx] [PATCH v3 09/16] drm/i915/guc: Convert guc_ads_private_data_reset " Lucas De Marchi
2022-02-18 20:36 ` Matthew Brost
2022-02-16 17:41 ` [Intel-gfx] [PATCH v3 10/16] drm/i915/guc: Convert golden context prep " Lucas De Marchi
2022-02-18 22:06 ` Matthew Brost
2022-02-16 17:41 ` [Intel-gfx] [PATCH v3 11/16] drm/i915/guc: Replace check for golden context size Lucas De Marchi
2022-02-18 20:39 ` Matthew Brost
2022-02-16 17:41 ` [Intel-gfx] [PATCH v3 12/16] drm/i915/guc: Convert mapping table to iosys_map Lucas De Marchi
2022-02-18 20:45 ` Matthew Brost
2022-02-16 17:41 ` [Intel-gfx] [PATCH v3 13/16] drm/i915/guc: Convert capture list " Lucas De Marchi
2022-02-18 20:50 ` Matthew Brost
2022-02-16 17:41 ` [Intel-gfx] [PATCH v3 14/16] drm/i915/guc: Convert guc_mmio_reg_state_init " Lucas De Marchi
2022-02-18 20:56 ` Matthew Brost
2022-02-16 17:41 ` [Intel-gfx] [PATCH v3 15/16] drm/i915/guc: Convert __guc_ads_init " Lucas De Marchi
2022-02-18 21:07 ` Matthew Brost
2022-02-16 17:41 ` [Intel-gfx] [PATCH v3 16/16] drm/i915/guc: Remove plain ads_blob pointer Lucas De Marchi
2022-02-18 21:12 ` Matthew Brost
2022-02-17 7:24 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for drm/i915/guc: Refactor ADS access to use iosys_map (rev3) Patchwork
2022-02-17 7:26 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2022-02-17 7:56 ` [Intel-gfx] ✗ Fi.CI.BAT: failure " Patchwork
2022-02-18 19:44 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for drm/i915/guc: Refactor ADS access to use iosys_map (rev4) Patchwork
2022-02-18 19:46 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2022-02-18 20:13 ` [Intel-gfx] ✓ Fi.CI.BAT: success " Patchwork
2022-02-19 12:47 ` [Intel-gfx] ✗ Fi.CI.IGT: failure " Patchwork
2022-02-19 18:46 ` Lucas De Marchi
2022-02-22 16:50 ` Vudum, Lakshminarayana
2022-02-22 16:46 ` [Intel-gfx] ✓ Fi.CI.IGT: success " 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=20220216174147.3073235-1-lucas.demarchi@intel.com \
--to=lucas.demarchi@intel.com \
--cc=christian.koenig@amd.com \
--cc=dri-devel@lists.freedesktop.org \
--cc=intel-gfx@lists.freedesktop.org \
--cc=sumit.semwal@linaro.org \
--cc=thomas.hellstrom@linux.intel.com \
--cc=tzimmermann@suse.de \
/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