<!DOCTYPE html><html><head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body>
<p><br>
</p>
<div class="moz-cite-prefix">On 4/29/2024 18:42, Lucas De Marchi
wrote:<br>
</div>
<blockquote type="cite" cite="mid:a3c4pszai2ysc74yod3yj7ulridwqrztotyu6lrtm3n2f3ghgu@y6lh66mlz2xe">On
Mon, Apr 29, 2024 at 09:14:30AM GMT, Chauhan, Shekhar wrote:
<br>
<blockquote type="cite">
<br>
On 4/27/2024 19:23, Lucas De Marchi wrote:
<br>
<blockquote type="cite">In order to detect duplicate
implementations for the same workaround,
<br>
early in the implementation of RTP it was decided to error out
even if
<br>
the values set are exactly the same. With the introduction of
18034896535
<br>
in commit 74671d23ca18 ("drm/xe/xe2: Add workaround
18034896535"), LNL
<br>
stepping with graphics stepping A1 now gives the following
error on
<br>
module load:
<br>
<br>
xe 0000:00:02.0: [drm] *ERROR* GT0: [GT OTHER] \
<br>
discarding save-restore reg e48c (clear: 00000200, set:
00000200,\
<br>
masked: yes, mcr: yes): ret=-22
<br>
<br>
RTP may be improved in the future, but for now simply join the
entries
<br>
like done with e.g. "1607297627, 1607030317, 1607186500".
<br>
<br>
Fixes: 74671d23ca18 ("drm/xe/xe2: Add workaround 18034896535")
<br>
Cc: Bommu Krishnaiah <a class="moz-txt-link-rfc2396E" href="mailto:krishnaiah.bommu@intel.com"><krishnaiah.bommu@intel.com></a>
<br>
Cc: Tejas Upadhyay <a class="moz-txt-link-rfc2396E" href="mailto:tejas.upadhyay@intel.com"><tejas.upadhyay@intel.com></a>
<br>
Cc: Matt Roper <a class="moz-txt-link-rfc2396E" href="mailto:matthew.d.roper@intel.com"><matthew.d.roper@intel.com></a>
<br>
Signed-off-by: Lucas De Marchi
<a class="moz-txt-link-rfc2396E" href="mailto:lucas.demarchi@intel.com"><lucas.demarchi@intel.com></a>
<br>
---
<br>
drivers/gpu/drm/xe/xe_wa.c | 7 +------
<br>
1 file changed, 1 insertion(+), 6 deletions(-)
<br>
<br>
diff --git a/drivers/gpu/drm/xe/xe_wa.c
b/drivers/gpu/drm/xe/xe_wa.c
<br>
index 9d9b7fa7a8f0..db7c7c7875c5 100644
<br>
--- a/drivers/gpu/drm/xe/xe_wa.c
<br>
+++ b/drivers/gpu/drm/xe/xe_wa.c
<br>
@@ -449,12 +449,7 @@ static const struct xe_rtp_entry_sr
engine_was[] = {
<br>
XE_RTP_RULES(GRAPHICS_VERSION(2004),
FUNC(xe_rtp_match_first_render_or_compute)),
<br>
XE_RTP_ACTIONS(SET(ROW_CHICKEN3,
XE2_EUPEND_CHK_FLUSH_DIS))
<br>
},
<br>
- { XE_RTP_NAME("16021540221"),
<br>
- XE_RTP_RULES(GRAPHICS_VERSION(2004), GRAPHICS_STEP(A0,
B0),
<br>
- FUNC(xe_rtp_match_first_render_or_compute)),
<br>
- XE_RTP_ACTIONS(SET(ROW_CHICKEN4, DISABLE_TDL_PUSH))
<br>
- },
<br>
- { XE_RTP_NAME("18034896535"),
<br>
+ { XE_RTP_NAME("18034896535, 16021540221"), /*
16021540221: GRAPHICS_STEP(A0, B0) */
<br>
XE_RTP_RULES(GRAPHICS_VERSION_RANGE(2001, 2004),
<br>
</blockquote>
<br>
w/a 16021540221 is applicable to only Steppings A0 to B0 (B0 not
inclusive). With your change, I think the w/a will be applied to
all the steppings. Wouldn't this be incorrect? Please provide
some guidance.
<br>
</blockquote>
<br>
both 18034896535 and 16021540221 have the same action:
<br>
<br>
SET(ROW_CHICKEN4, DISABLE_TDL_PUSH)
<br>
<br>
In other words, 18034896535 applies to a superset of platforms
<br>
16021540221 applies to. So even when we remove support for
steppings A*,
<br>
we want to keep the WA (possibly removing the comment about
16021540221
<br>
and renaming it).
<br>
</blockquote>
<p>Okay, this makes sense. Thanks. <br>
</p>
<p><i>-shekhar</i><br>
</p>
<blockquote type="cite" cite="mid:a3c4pszai2ysc74yod3yj7ulridwqrztotyu6lrtm3n2f3ghgu@y6lh66mlz2xe">
<br>
Lucas De Marchi
<br>
<br>
<blockquote type="cite">
<br>
<blockquote type="cite">
FUNC(xe_rtp_match_first_render_or_compute)),
<br>
XE_RTP_ACTIONS(SET(ROW_CHICKEN4, DISABLE_TDL_PUSH))
<br>
</blockquote>
<br>
-- <br>
-shekhar
<br>
<br>
</blockquote>
</blockquote>
<pre class="moz-signature" cols="72">--
-shekhar</pre>
</body>
</html>