drm/amd/display: use div_s64() for 64-bit division
Subsystems (3)
PATCH 1/1 drm/amd/display: use div_s64() for 64-bit division
From: Arnd Bergmann <[email protected]>
The open-coded 64-bit division causes a link error on 32-bit machines:
ERROR: modpost: "__udivdi3" [drivers/gpu/drm/amd/amdgpu/amdgpu.ko] undefined!
ERROR: modpost: "__divdi3" [drivers/gpu/drm/amd/amdgpu/amdgpu.ko] undefined!
Use the div_s64() to perform the division here. One of them was an unsigned division originally, but it looks like signed division was intended, so use that to consistently allow a negative delay.
--- drivers/gpu/drm/amd/display/dc/dcn30/dcn30_hwseq.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)
drivers/gpu/drm/amd/display/dc/dcn30/dcn30_hwseq.c +6 -6
@@ -772,8 +772,8 @@ bool dcn30_apply_idle_power_optimizations(struct dc *dc, bool enable) cursor_cache_enable ? &cursor_attr : NULL)) {unsigned int v_total = stream->adjust.v_total_max ? stream->adjust.v_total_max : stream->timing.v_total; -unsigned int refresh_hz = (unsigned long long ) stream->timing.pix_clk_100hz * -100LL / (v_total * stream->timing.h_total); +unsigned int refresh_hz =div_s64 ((unsigned long long ) stream->timing.pix_clk_100hz * +100LL , v_total * stream->timing.h_total);/* * one frame time in microsec:
@@ -800,8 +800,8 @@ bool dcn30_apply_idle_power_optimizations(struct dc *dc, bool enable)unsigned int denom = refresh_hz *6528 ;unsigned int stutter_period = dc->current_state->perf_params.stutter_period_us; - tmr_delay = (((1000000LL +2 * stutter_period * refresh_hz) * - (100LL + dc->debug.mall_additional_timer_percent) + denom -1 ) / + tmr_delay =div_s64 (((1000000LL +2 * stutter_period * refresh_hz) * + (100LL + dc->debug.mall_additional_timer_percent) + denom -1 ), denom) -64LL ;/* scale should be increased until it fits into 6 bits */
@@ -815,8 +815,8 @@ bool dcn30_apply_idle_power_optimizations(struct dc *dc, bool enable) } denom *=2 ; - tmr_delay = (((1000000LL +2 * stutter_period * refresh_hz) * - (100LL + dc->debug.mall_additional_timer_percent) + denom -1 ) / + tmr_delay =div_s64 (((1000000LL +2 * stutter_period * refresh_hz) * + (100LL + dc->debug.mall_additional_timer_percent) + denom -1 ), denom) -64LL ; }
Signature · 6 lines
2.29.2 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
[AMD Public Use]
Hi Arnd Bergmann,
Thanks for your patch. This link error during compile has been fixed by below commit and been submitted to drm-next branch already.
5da047444e82 drm/amd/display: fix 64-bit division issue on 32-bit OS
Regards,
Guchun
-----Original Message-----
From: amd-gfx <[email protected]> On Behalf Of Arnd Bergmann
Sent: Monday, January 25, 2021 7:40 PM
To: Wentland, Harry <[email protected]>; Li, Sun peng (Leo) <[email protected]>; Deucher, Alexander <[email protected]>; Koenig, Christian <[email protected]>; David Airlie <[email protected]>; Daniel Vetter <[email protected]>; Aberback, Joshua <[email protected]>; Lakha, Bhawanpreet <[email protected]>; Kazlauskas, Nicholas <[email protected]>
Cc: Arnd Bergmann <[email protected]>; Chalmers, Wesley <[email protected]>; Zhuo, Qingqing <[email protected]>; Siqueira, Rodrigo <[email protected]>; [email protected]; [email protected]; [email protected]; Jacky Liao <[email protected]>; Leung, Martin <[email protected]>
Subject: [PATCH] drm/amd/display: use div_s64() for 64-bit division
From: Arnd Bergmann <[email protected]>
The open-coded 64-bit division causes a link error on 32-bit machines:
ERROR: modpost: "__udivdi3" [drivers/gpu/drm/amd/amdgpu/amdgpu.ko] undefined!
ERROR: modpost: "__divdi3" [drivers/gpu/drm/amd/amdgpu/amdgpu.ko] undefined!
Use the div_s64() to perform the division here. One of them was an unsigned division originally, but it looks like signed division was intended, so use that to consistently allow a negative delay.
--- drivers/gpu/drm/amd/display/dc/dcn30/dcn30_hwseq.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)
drivers/gpu/drm/amd/display/dc/dcn30/dcn30_hwseq.c +4 -4
@@ -772,8 +772,8 @@ bool dcn30_apply_idle_power_optimizations(struct dc *dc, bool enable) cursor_cache_enable ? &cursor_attr : NULL)) {unsigned int v_total = stream->adjust.v_total_max ? stream->adjust.v_total_max : stream->timing.v_total; -unsigned int refresh_hz = (unsigned long long ) stream->timing.pix_clk_100hz * -100LL / (v_total * stream->timing.h_total); +unsigned int refresh_hz =div_s64 ((unsigned long long ) stream->timing.pix_clk_100hz * +100LL , v_total * stream->timing.h_total);/* * one frame time in microsec:
@@ -800,8 +800,8 @@ bool dcn30_apply_idle_power_optimizations(struct dc *dc, bool enable)unsigned int denom = refresh_hz *6528 ;unsigned int stutter_period = dc->current_state->perf_params.stutter_period_us; - tmr_delay = (((1000000LL +2 * stutter_period * refresh_hz) * - (100LL + dc->debug.mall_additional_timer_percent) + denom -1 ) / + tmr_delay =div_s64 (((1000000LL +2 * stutter_period * refresh_hz) * + (100LL + dc->debug.mall_additional_timer_percent) + denom -1 ), denom) -64LL ; /* scale should be increased until it fits into 6 bits */ @@ -815,8 +815,8 @@ bool dcn30_apply_idle_power_optimizations(struct dc *dc, bool enable)
}
denom *= 2;
- tmr_delay = (((1000000LL + 2 * stutter_period * refresh_hz) *
- (100LL + dc->debug.mall_additional_timer_percent) + denom - 1) /
+ tmr_delay = div_s64(((1000000LL + 2 * stutter_period * refresh_hz) *
+ (100LL + dc->debug.mall_additional_timer_percent) + denom - 1),
denom) - 64LL;
}
--
2.29.2
_______________________________________________
amd-gfx mailing list
[email protected]
https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.freedesktop.org%2Fmailman%2Flistinfo%2Famd-gfx&data=04%7C01%7Cguchun.chen%40amd.com%7C4bb97aae9edc4153392c08d8c1260048%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637471716255231899%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=kLdkVHfkYx%2Bd249%2BmtG5GJTq295Pxzw7mgTe0FD8QvY%3D&reserved=0
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Quoted 8 lines · Guchun
[AMD Public Use]
Hi Arnd Bergmann,
Thanks for your patch. This link error during compile has been fixed by below commit and been submitted to drm-next branch already.
5da047444e82drm/amd/display: fix 64-bit division issue on 32-bit OS
Ok, thanks. I assume this will make it into linux-next in the coming days then.
Arnd
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel