From: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
To: "Zbigniew Kempczyński" <zbigniew.kempczynski@intel.com>,
igt-dev@lists.freedesktop.org
Subject: Re: [PATCH i-g-t 03/12] lib/intel_bufops: Add linear-to-none copy path
Date: Mon, 17 Jun 2024 14:16:20 +0300 [thread overview]
Message-ID: <f958a0bd-4c4d-4e6e-96f0-c1bf70763c24@gmail.com> (raw)
In-Reply-To: <20240527073345.54729-4-zbigniew.kempczynski@intel.com>
On 27.5.2024 10.33, Zbigniew Kempczyński wrote:
> At first glance copying linear to linear (none) is not necessary
> as ordinary memcpy() may be used. But adding this makes iterating
> over all possible tilings much easier.
>
> In the upcoming patch I'm going to introduce tiling detection tool
> which iterates over all tilings where linear copy is also exercised.
>
> Signed-off-by: Zbigniew Kempczyński <zbigniew.kempczynski@intel.com>
> Cc: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
> ---
> lib/intel_bufops.c | 35 ++++++++++++++++++++++++++++++++++-
> 1 file changed, 34 insertions(+), 1 deletion(-)
>
> diff --git a/lib/intel_bufops.c b/lib/intel_bufops.c
> index b4ccf4c093..30ba2547dc 100644
> --- a/lib/intel_bufops.c
> +++ b/lib/intel_bufops.c
> @@ -292,6 +292,17 @@ static unsigned long swizzle_addr(void *ptr, uint32_t swizzle)
> }
> }
>
> +static void *linear_ptr(void *ptr,
> + unsigned int x, unsigned int y,
> + unsigned int stride, unsigned int cpp)
> +{
> + int pos;
> +
> + pos = (stride/cpp * y + x) * cpp;
I was for a while thinking about this .. but I'd expect gcc will be able
to take out that division and on source level we can keep this as is.
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
> +
> + return ptr + pos;
> +}
> +
> static void *x_ptr(void *ptr,
> unsigned int x, unsigned int y,
> unsigned int stride, unsigned int cpp)
> @@ -418,6 +429,9 @@ static tile_fn __get_tile_fn_ptr(int tiling)
> tile_fn fn = NULL;
>
> switch (tiling) {
> + case I915_TILING_NONE:
> + fn = linear_ptr;
> + break;
> case I915_TILING_X:
> fn = x_ptr;
> break;
> @@ -598,6 +612,13 @@ static void __copy_linear_to(int fd, struct intel_buf *buf,
> munmap(map, buf->surface[0].size);
> }
>
> +static void copy_linear_to_none(struct buf_ops *bops, struct intel_buf *buf,
> + uint32_t *linear)
> +{
> + DEBUGFN();
> + __copy_linear_to(bops->fd, buf, linear, I915_TILING_NONE, 0);
> +}
> +
> static void copy_linear_to_x(struct buf_ops *bops, struct intel_buf *buf,
> uint32_t *linear)
> {
> @@ -655,6 +676,13 @@ static void __copy_to_linear(int fd, struct intel_buf *buf,
> munmap(map, buf->surface[0].size);
> }
>
> +static void copy_none_to_linear(struct buf_ops *bops, struct intel_buf *buf,
> + uint32_t *linear)
> +{
> + DEBUGFN();
> + __copy_to_linear(bops->fd, buf, linear, I915_TILING_NONE, 0);
> +}
> +
> static void copy_x_to_linear(struct buf_ops *bops, struct intel_buf *buf,
> uint32_t *linear)
> {
> @@ -1653,13 +1681,14 @@ static struct buf_ops *__buf_ops_create(int fd, bool check_idempotency)
> bops->driver == INTEL_DRIVER_I915 ? "i915" : "xe");
>
> if (bops->driver == INTEL_DRIVER_XE) {
> + bops->linear_to = copy_linear_to_none;
> + bops->to_linear = copy_none_to_linear;
> bops->linear_to_x = copy_linear_to_x;
> bops->x_to_linear = copy_x_to_linear;
> bops->linear_to_y = copy_linear_to_y;
> bops->y_to_linear = copy_y_to_linear;
> bops->linear_to_tile4 = copy_linear_to_tile4;
> bops->tile4_to_linear = copy_tile4_to_linear;
> -
> bops->linear_to_yf = NULL;
> bops->yf_to_linear = NULL;
> bops->linear_to_ys = NULL;
> @@ -1869,6 +1898,10 @@ bool buf_ops_set_software_tiling(struct buf_ops *bops,
> }
>
> switch (tiling) {
> + case I915_TILING_NONE:
> + igt_debug("-> use SW on tiling NONE\n");
> + break;
> +
> case I915_TILING_X:
> if (use_software_tiling) {
> bool supported = buf_ops_has_tiling_support(bops, tiling);
next prev parent reply other threads:[~2024-06-17 11:16 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-05-27 7:33 [PATCH i-g-t 00/12] Introduce intel_tiling_detect tool Zbigniew Kempczyński
2024-05-27 7:33 ` [PATCH i-g-t 01/12] lib/intel_bufops: Fix offset returned for Tile4 Zbigniew Kempczyński
2024-06-03 7:26 ` Juha-Pekka Heikkila
2024-05-27 7:33 ` [PATCH i-g-t 02/12] lib/intel_bufops: Fix mapping/unmapping for i915 and xe Zbigniew Kempczyński
2024-06-03 7:31 ` Juha-Pekka Heikkila
2024-05-27 7:33 ` [PATCH i-g-t 03/12] lib/intel_bufops: Add linear-to-none copy path Zbigniew Kempczyński
2024-06-17 11:16 ` Juha-Pekka Heikkila [this message]
2024-05-27 7:33 ` [PATCH i-g-t 04/12] lib/intel_bufops: Add Tile4 in linear_to and to_linear helpers Zbigniew Kempczyński
2024-06-17 11:16 ` Juha-Pekka Heikkila
2024-05-27 7:33 ` [PATCH i-g-t 05/12] lib/intel_bufops: Add Yf tiling Zbigniew Kempczyński
2024-06-17 11:16 ` Juha-Pekka Heikkila
2024-05-27 7:33 ` [PATCH i-g-t 06/12] lib/intel_blt: Add Yf tiling in block-copy Zbigniew Kempczyński
2024-06-17 11:16 ` Juha-Pekka Heikkila
2024-05-27 7:33 ` [PATCH i-g-t 07/12] lib/intel_bufops: Add Ys tiling in linear_to and to_linear path Zbigniew Kempczyński
2024-06-17 11:16 ` Juha-Pekka Heikkila
2024-05-27 7:33 ` [PATCH i-g-t 08/12] lib/intel_bufops: Drop unnecessary Tile4 function assignment Zbigniew Kempczyński
2024-06-17 11:16 ` Juha-Pekka Heikkila
2024-05-27 7:33 ` [PATCH i-g-t 09/12] lib/intel_bufops: Preparation for adding software tiling for Tile64 Zbigniew Kempczyński
2024-06-17 11:17 ` Juha-Pekka Heikkila
2024-05-27 7:33 ` [PATCH i-g-t 10/12] lib/intel_bufops: Drop Tile4 swizzling Zbigniew Kempczyński
2024-06-17 11:17 ` Juha-Pekka Heikkila
2024-05-27 7:33 ` [PATCH i-g-t 11/12] lib/intel_bufops: Don't disable x and y on platforms without swizzling Zbigniew Kempczyński
2024-06-17 16:43 ` Juha-Pekka Heikkila
2024-05-27 7:33 ` [PATCH i-g-t 12/12] tools/intel_tiling_detect: Introduce tiling detection tool Zbigniew Kempczyński
2024-06-17 19:21 ` Juha-Pekka Heikkila
2024-06-18 4:51 ` Zbigniew Kempczyński
2024-06-17 19:46 ` Kamil Konieczny
2024-06-18 5:15 ` Zbigniew Kempczyński
2024-05-27 18:40 ` ✓ Fi.CI.BAT: success for Introduce intel_tiling_detect tool Patchwork
2024-05-27 18:45 ` ✓ CI.xeBAT: " Patchwork
2024-05-28 7:29 ` ✗ Fi.CI.IGT: failure " Patchwork
2024-05-28 7:46 ` Zbigniew Kempczyński
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=f958a0bd-4c4d-4e6e-96f0-c1bf70763c24@gmail.com \
--to=juhapekka.heikkila@gmail.com \
--cc=igt-dev@lists.freedesktop.org \
--cc=zbigniew.kempczynski@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