drm/i915/dp: reduce link M/N parameters
Subsystems (2)
PATCH 1/1 drm/i915/dp: reduce link M/N parameters
Several major vendor USB-C->HDMI converters, in particular the DA200, fail to recover a 5.4 GHz 1 lane signal if the link N is greater than 0x80000.
The link M and N depend on the pixel clock and link clock ratio. With current code link N exceeds 0x80000 only when link clock >= 540000 kHz. Except for the eDP intermediate link clocks, at least the four least significant bits are always zero. Just one bit shift right would be enough to bring even the DP 1.4 810000 kHz link clock under 0x80000 link N. The pixel clock for modes that require a link clock >= 540000 kHz would also have several least significant bits zero. Unless the user provides a mode with an odd pixel clock value, we can reduce the numbers to reach the goal, with no loss in precision.
The DP spec even mentions sources making choices that "allow for static and relatively small Mvid and Nvid values", thus reducing the link M/N regardless of the sink in question seems justified.
Everything here is based on the work and information gathered by Clint Taylor <[email protected]>. This is just an iteration to reduce the parameters regardless of lane count, link rate, or sink.
Reference: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93578
---
This is cc: stable material, but due to the slight risk of regressions (there's always the risk, however small, when you change parameters that affect all sinks) I'd prefer letting this simmer for a while, and asking for an explicit stable backport afterwards.
--- drivers/gpu/drm/i915/intel_display.c | 9 +++++++++ 1 file changed, 9 insertions(+)
drivers/gpu/drm/i915/intel_display.c +9 -0
@@ -6337,6 +6337,15 @@ intel_reduce_m_n_ratio(uint32_t *num, uint32_t *den)static void compute_m_n (unsigned int m,unsigned int n,uint32_t *ret_m,uint32_t *ret_n) { +/* +* Reduce M/N as much as possible without loss in precision. Several DP +* dongles in particular seem to be fussy about too large M/N values. +*/ +while ((m &1 ) ==0 && (n &1 ) ==0 ) { + m >>=1 ; + n >>=1 ; + } + *ret_n = min_t(unsigned int, roundup_pow_of_two(n), DATA_LINK_N_MAX); *ret_m =div_u64 ((uint64_t ) m * *ret_n, n);intel_reduce_m_n_ratio (ret_m, ret_n);
Signature · 6 lines
2.1.4 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✗ Fi.CI.BAT: failure for drm/i915/dp: reduce link M/N parameters
== Series Details ==
== Series Details ==
Series: drm/i915/dp: reduce link M/N parameters
URL : https://patchwork.freedesktop.org/series/21927/
State : failure
== Summary ==
Series 21927v1 drm/i915/dp: reduce link M/N parameters
https://patchwork.freedesktop.org/api/1.0/series/21927/revisions/1/mbox/
Test gem_exec_fence:
Subgroup nb-await-default:
pass -> INCOMPLETE (fi-hsw-4770)
Test gem_exec_flush:
Subgroup basic-batch-kernel-default-uc:
fail -> PASS (fi-snb-2600) fdo#100007fdo#100007 https://bugs.freedesktop.org/show_bug.cgi?id=100007
fi-bdw-5557u total:278 pass:267 dwarn:0 dfail:0 fail:0 skip:11 time: 462s fi-bdw-gvtdvm total:278 pass:256 dwarn:8 dfail:0 fail:0 skip:14 time: 458s fi-bsw-n3050 total:278 pass:239 dwarn:0 dfail:0 fail:0 skip:39 time: 582s fi-bxt-j4205 total:278 pass:259 dwarn:0 dfail:0 fail:0 skip:19 time: 538s fi-bxt-t5700 total:278 pass:258 dwarn:0 dfail:0 fail:0 skip:20 time: 596s fi-byt-j1900 total:278 pass:251 dwarn:0 dfail:0 fail:0 skip:27 time: 512s fi-byt-n2820 total:278 pass:247 dwarn:0 dfail:0 fail:0 skip:31 time: 511s fi-hsw-4770 total:49 pass:42 dwarn:0 dfail:0 fail:0 skip:6 time: 0s fi-hsw-4770r total:278 pass:262 dwarn:0 dfail:0 fail:0 skip:16 time: 427s fi-ilk-650 total:278 pass:228 dwarn:0 dfail:0 fail:0 skip:50 time: 436s fi-ivb-3520m total:278 pass:260 dwarn:0 dfail:0 fail:0 skip:18 time: 518s fi-ivb-3770 total:278 pass:260 dwarn:0 dfail:0 fail:0 skip:18 time: 493s fi-kbl-7500u total:278 pass:260 dwarn:0 dfail:0 fail:0 skip:18 time: 489s fi-skl-6260u total:278 pass:268 dwarn:0 dfail:0 fail:0 skip:10 time: 485s fi-skl-6700hq total:278 pass:261 dwarn:0 dfail:0 fail:0 skip:17 time: 605s fi-skl-6700k total:278 pass:256 dwarn:4 dfail:0 fail:0 skip:18 time: 486s fi-skl-6770hq total:278 pass:268 dwarn:0 dfail:0 fail:0 skip:10 time: 522s fi-skl-gvtdvm total:278 pass:265 dwarn:0 dfail:0 fail:0 skip:13 time: 457s fi-snb-2520m total:278 pass:250 dwarn:0 dfail:0 fail:0 skip:28 time: 546s fi-snb-2600 total:278 pass:249 dwarn:0 dfail:0 fail:0 skip:29 time: 417s
71f45aec6fccfe64434836abbea0a10cb3b71755 drm-tip: 2017y-03m-27d-08h-58m-18s UTC integration manifest b79182e drm/i915/dp: reduce link M/N parameters
== Logs ==
For more details see: https://intel-gfx-ci.01.org/CI/Patchwork_4313/
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 55 lines · Jani Nikula
Several major vendor USB-C->HDMI converters, in particular the DA200, fail to recover a 5.4 GHz 1 lane signal if the link N is greater than
0x80000.The link M and N depend on the pixel clock and link clock ratio. With current code link N exceeds
0x80000only when link clock >= 540000 kHz. Except for the eDP intermediate link clocks, at least the four least significant bits are always zero. Just one bit shift right would be enough to bring even the DP 1.4 810000 kHz link clock under0x80000link N. The pixel clock for modes that require a link clock >= 540000 kHz would also have several least significant bits zero. Unless the user provides a mode with an odd pixel clock value, we can reduce the numbers to reach the goal, with no loss in precision.The DP spec even mentions sources making choices that "allow for static and relatively small Mvid and Nvid values", thus reducing the link M/N regardless of the sink in question seems justified.
Everything here is based on the work and information gathered by Clint Taylor <[email protected]>. This is just an iteration to reduce the parameters regardless of lane count, link rate, or sink.
Reference: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93578Tested-by Mads <[email protected]>Tested-by PJ <[email protected]>Tested-by François Guerraz <[email protected]>Tested-by Lev Popov <[email protected]>Tested-by Igor Krivenko <[email protected]>Cc Clint Taylor <[email protected]>Cc Anusha Srivatsa <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Daniel Vetter <[email protected]>Signed-off-by Jani Nikula <[email protected]>---
This is cc: stable material, but due to the slight risk of regressions (there's always the risk, however small, when you change parameters that affect all sinks) I'd prefer letting this simmer for a while, and asking for an explicit stable backport afterwards.
--- drivers/gpu/drm/i915/intel_display.c | 9 +++++++++ 1 file changed, 9 insertions(+)drivers/gpu/drm/i915/intel_display.c +2
-0@@ -6337,6 +6337,15 @@ intel_reduce_m_n_ratio(uint32_t *num, uint32_t *den)static void compute_m_n (unsigned int m,unsigned int n,uint32_t *ret_m,uint32_t *ret_n) { +/* +* Reduce M/N as much as possible without loss in precision. Several DP
drivers/gpu/drm/i915/intel_display.c intel_reduce_m_n_ratio(uint32_t *num, uint32_t *den) · line 6351
Quoted 10 lines
drivers/gpu/drm/i915/intel_display.c +7
-0+* dongles in particular seem to be fussy about too large M/N values. +*/ +while ((m &1 ) ==0 && (n &1 ) ==0 ) { + m >>=1 ; + n >>=1 ; + } + *ret_n = min_t(unsigned int, roundup_pow_of_two(n), DATA_LINK_N_MAX); *ret_m =div_u64 ((uint64_t ) m * *ret_n, n);intel_reduce_m_n_ratio (ret_m, ret_n);
Can't we push this into reduce_m_n_ratio?
-Daniel
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 60 lines · Daniel Vetter
Quoted 59 lines · Jani Nikula
Several major vendor USB-C->HDMI converters, in particular the DA200, fail to recover a 5.4 GHz 1 lane signal if the link N is greater than
0x80000.The link M and N depend on the pixel clock and link clock ratio. With current code link N exceeds
0x80000only when link clock >= 540000 kHz. Except for the eDP intermediate link clocks, at least the four least significant bits are always zero. Just one bit shift right would be enough to bring even the DP 1.4 810000 kHz link clock under0x80000link N. The pixel clock for modes that require a link clock >= 540000 kHz would also have several least significant bits zero. Unless the user provides a mode with an odd pixel clock value, we can reduce the numbers to reach the goal, with no loss in precision.The DP spec even mentions sources making choices that "allow for static and relatively small Mvid and Nvid values", thus reducing the link M/N regardless of the sink in question seems justified.
Everything here is based on the work and information gathered by Clint Taylor <[email protected]>. This is just an iteration to reduce the parameters regardless of lane count, link rate, or sink.
Reference: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93578Tested-by Mads <[email protected]>Tested-by PJ <[email protected]>Tested-by François Guerraz <[email protected]>Tested-by Lev Popov <[email protected]>Tested-by Igor Krivenko <[email protected]>Cc Clint Taylor <[email protected]>Cc Anusha Srivatsa <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Daniel Vetter <[email protected]>Signed-off-by Jani Nikula <[email protected]>---
This is cc: stable material, but due to the slight risk of regressions (there's always the risk, however small, when you change parameters that affect all sinks) I'd prefer letting this simmer for a while, and asking for an explicit stable backport afterwards.
--- drivers/gpu/drm/i915/intel_display.c | 9 +++++++++ 1 file changed, 9 insertions(+)drivers/gpu/drm/i915/intel_display.c +6
-0@@ -6337,6 +6337,15 @@ intel_reduce_m_n_ratio(uint32_t *num, uint32_t *den)static void compute_m_n (unsigned int m,unsigned int n,uint32_t *ret_m,uint32_t *ret_n) { +/* +* Reduce M/N as much as possible without loss in precision. Several DP +* dongles in particular seem to be fussy about too large M/N values. +*/ +while ((m &1 ) ==0 && (n &1 ) ==0 ) { + m >>=1 ;
Quoted 8 lines
drivers/gpu/drm/i915/intel_display.c intel_reduce_m_n_ratio(uint32_t *num, uint32_t *den) · line 6351
Quoted 6 lines
drivers/gpu/drm/i915/intel_display.c +3
-0+ n >>=1 ; + } + *ret_n = min_t(unsigned int, roundup_pow_of_two(n), DATA_LINK_N_MAX); *ret_m =div_u64 ((uint64_t ) m * *ret_n, n);intel_reduce_m_n_ratio (ret_m, ret_n);Can't we push this into
reduce_m_n_ratio?
After *ret_m = div_u64((uint64_t) m * *ret_n, n); it's not guaranteed that we could shift at all. The reduction needs to be done on the original m, n being passed in.
BR,
Jani.
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 70 lines · Jani Nikula
Quoted 60 lines · Daniel Vetter
Quoted 59 lines · Jani Nikula
Several major vendor USB-C->HDMI converters, in particular the DA200, fail to recover a 5.4 GHz 1 lane signal if the link N is greater than
0x80000.The link M and N depend on the pixel clock and link clock ratio. With current code link N exceeds
0x80000only when link clock >= 540000 kHz. Except for the eDP intermediate link clocks, at least the four least significant bits are always zero. Just one bit shift right would be enough to bring even the DP 1.4 810000 kHz link clock under0x80000link N. The pixel clock for modes that require a link clock >= 540000 kHz would also have several least significant bits zero. Unless the user provides a mode with an odd pixel clock value, we can reduce the numbers to reach the goal, with no loss in precision.The DP spec even mentions sources making choices that "allow for static and relatively small Mvid and Nvid values", thus reducing the link M/N regardless of the sink in question seems justified.
Everything here is based on the work and information gathered by Clint Taylor <[email protected]>. This is just an iteration to reduce the parameters regardless of lane count, link rate, or sink.
Reference: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93578Tested-by Mads <[email protected]>Tested-by PJ <[email protected]>Tested-by François Guerraz <[email protected]>Tested-by Lev Popov <[email protected]>Tested-by Igor Krivenko <[email protected]>Cc Clint Taylor <[email protected]>Cc Anusha Srivatsa <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Daniel Vetter <[email protected]>Signed-off-by Jani Nikula <[email protected]>---
This is cc: stable material, but due to the slight risk of regressions (there's always the risk, however small, when you change parameters that affect all sinks) I'd prefer letting this simmer for a while, and asking for an explicit stable backport afterwards.
--- drivers/gpu/drm/i915/intel_display.c | 9 +++++++++ 1 file changed, 9 insertions(+)drivers/gpu/drm/i915/intel_display.c +6
-0@@ -6337,6 +6337,15 @@ intel_reduce_m_n_ratio(uint32_t *num, uint32_t *den)static void compute_m_n (unsigned int m,unsigned int n,uint32_t *ret_m,uint32_t *ret_n) { +/* +* Reduce M/N as much as possible without loss in precision. Several DP +* dongles in particular seem to be fussy about too large M/N values. +*/ +while ((m &1 ) ==0 && (n &1 ) ==0 ) { + m >>=1 ;Quoted 8 lines
drivers/gpu/drm/i915/intel_display.c intel_reduce_m_n_ratio(uint32_t *num, uint32_t *den) · line 6351
Quoted 6 lines
drivers/gpu/drm/i915/intel_display.c +3
-0+ n >>=1 ; + } + *ret_n = min_t(unsigned int, roundup_pow_of_two(n), DATA_LINK_N_MAX); *ret_m =div_u64 ((uint64_t ) m * *ret_n, n);intel_reduce_m_n_ratio (ret_m, ret_n);Can't we push this into
reduce_m_n_ratio?
Quoted 4 lines
After
*ret_m=div_u64((uint64_t) m * *ret_n, n); it's not guaranteed that we could shift at all. The reduction needs to be done on the original m, n being passed in.
Quoted 4 lines
BR,
Jani.
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Jani Nikula
AppliedQuoted 71 lines · Clint Taylor
Quoted 70 lines · Jani Nikula
Quoted 68 lines · Daniel Vetter
Quoted 55 lines · Jani Nikula
Several major vendor USB-C->HDMI converters, in particular the DA200, fail to recover a 5.4 GHz 1 lane signal if the link N is greater than
0x80000.The link M and N depend on the pixel clock and link clock ratio. With current code link N exceeds
0x80000only when link clock >= 540000 kHz. Except for the eDP intermediate link clocks, at least the four least significant bits are always zero. Just one bit shift right would be enough to bring even the DP 1.4 810000 kHz link clock under0x80000link N. The pixel clock for modes that require a link clock >= 540000 kHz would also have several least significant bits zero. Unless the user provides a mode with an odd pixel clock value, we can reduce the numbers to reach the goal, with no loss in precision.The DP spec even mentions sources making choices that "allow for static and relatively small Mvid and Nvid values", thus reducing the link M/N regardless of the sink in question seems justified.
Everything here is based on the work and information gathered by Clint Taylor <[email protected]>. This is just an iteration to reduce the parameters regardless of lane count, link rate, or sink.
Reference: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93578Tested-by Mads <[email protected]>Tested-by PJ <[email protected]>Tested-by François Guerraz <[email protected]>Tested-by Lev Popov <[email protected]>Tested-by Igor Krivenko <[email protected]>Cc Clint Taylor <[email protected]>Cc Anusha Srivatsa <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Daniel Vetter <[email protected]>Signed-off-by Jani Nikula <[email protected]>---
This is cc: stable material, but due to the slight risk of regressions (there's always the risk, however small, when you change parameters that affect all sinks) I'd prefer letting this simmer for a while, and asking for an explicit stable backport afterwards.
--- drivers/gpu/drm/i915/intel_display.c | 9 +++++++++ 1 file changed, 9 insertions(+)drivers/gpu/drm/i915/intel_display.c +2
-0@@ -6337,6 +6337,15 @@ intel_reduce_m_n_ratio(uint32_t *num, uint32_t *den)static void compute_m_n (unsigned int m,unsigned int n,uint32_t *ret_m,uint32_t *ret_n) { +/* +* Reduce M/N as much as possible without loss in precision. Several DP drivers/gpu/drm/i915/intel_display.c intel_reduce_m_n_ratio(uint32_t *num, uint32_t *den) · line 6351
Quoted 10 lines
drivers/gpu/drm/i915/intel_display.c +7
-0+* dongles in particular seem to be fussy about too large M/N values. +*/ +while ((m &1 ) ==0 && (n &1 ) ==0 ) { + m >>=1 ; + n >>=1 ; + } + *ret_n = min_t(unsigned int, roundup_pow_of_two(n), DATA_LINK_N_MAX); *ret_m =div_u64 ((uint64_t ) m * *ret_n, n);intel_reduce_m_n_ratio (ret_m, ret_n);Can't we push this into
reduce_m_n_ratio?
Quoted 7 lines
Quoted 4 lines
After
*ret_m=div_u64((uint64_t) m * *ret_n, n); it's not guaranteed that we could shift at all. The reduction needs to be done on the original m, n being passed in.Tested-by Clint Taylor <[email protected]>Reviewed-by Clint Taylor <[email protected]>
Pushed to drm-intel-next-queued, thanks Clint for your work in getting at the roots of the problem, and everyone for all the review and testing!
BR,
Jani.
Quoted 6 lines
Quoted 5 lines
BR,
Jani.
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 10 lines · Jani Nikula
Several major vendor USB-C->HDMI converters, in particular the DA200, fail to recover a 5.4 GHz 1 lane signal if the link N is greater than
0x80000.The link M and N depend on the pixel clock and link clock ratio. With current code link N exceeds
0x80000only when link clock >= 540000 kHz. Except for the eDP intermediate link clocks, at least the four least significant bits are always zero. Just one bit shift right would be enough to bring even the DP 1.4 810000 kHz link clock under0x80000link N.
I don't understand this part, the right shift is applied to 'n' and not link_clock. For, link_clock=810000kHz, lane=1, n is 62E080h (810000*1*8) and right-shifting this by 1 does bring it under 80000h. Can you please clarify this?
The patch itself looks right and is well within the Spec.
Quoted 46 lines
The pixel clock for modes that require a link clock >= 540000 kHz would also have several least significant bits zero. Unless the user provides a mode with an odd pixel clock value, we can reduce the numbers to reach the goal, with no loss in precision.
The DP spec even mentions sources making choices that "allow for static and relatively small Mvid and Nvid values", thus reducing the link M/N regardless of the sink in question seems justified.
Everything here is based on the work and information gathered by Clint Taylor <[email protected]>. This is just an iteration to reduce the parameters regardless of lane count, link rate, or sink.
Reference: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93578Tested-by Mads <[email protected]>Tested-by PJ <[email protected]>Tested-by François Guerraz <[email protected]>Tested-by Lev Popov <[email protected]>Tested-by Igor Krivenko <[email protected]>Cc Clint Taylor <[email protected]>Cc Anusha Srivatsa <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Daniel Vetter <[email protected]>Signed-off-by Jani Nikula <[email protected]>---
This is cc: stable material, but due to the slight risk of regressions (there's always the risk, however small, when you change parameters that affect all sinks) I'd prefer letting this simmer for a while, and asking for an explicit stable backport afterwards.
--- drivers/gpu/drm/i915/intel_display.c | 9 +++++++++ 1 file changed, 9 insertions(+)drivers/gpu/drm/i915/intel_display.c +2
-0@@ -6337,6 +6337,15 @@ intel_reduce_m_n_ratio(uint32_t *num, uint32_t *den)static void compute_m_n (unsigned int m,unsigned int n,uint32_t *ret_m,uint32_t *ret_n) { +/* +* Reduce M/N as much as possible without loss in precision. Several DP
drivers/gpu/drm/i915/intel_display.c intel_reduce_m_n_ratio(uint32_t *num, uint32_t *den) · line 6351
Quoted 10 lines
drivers/gpu/drm/i915/intel_display.c +7
-0+* dongles in particular seem to be fussy about too large M/N values. +*/ +while ((m &1 ) ==0 && (n &1 ) ==0 ) { + m >>=1 ; + n >>=1 ; + } + *ret_n = min_t(unsigned int, roundup_pow_of_two(n), DATA_LINK_N_MAX); *ret_m =div_u64 ((uint64_t ) m * *ret_n, n);intel_reduce_m_n_ratio (ret_m, ret_n);
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 12 lines · Dhinakaran
Quoted 10 lines · Jani Nikula
Several major vendor USB-C->HDMI converters, in particular the DA200, fail to recover a 5.4 GHz 1 lane signal if the link N is greater than
0x80000.The link M and N depend on the pixel clock and link clock ratio. With current code link N exceeds
0x80000only when link clock >= 540000 kHz. Except for the eDP intermediate link clocks, at least the four least significant bits are always zero. Just one bit shift right would be enough to bring even the DP 1.4 810000 kHz link clock under0x80000link N.
Quoted 4 lines
I don't understand this part, the right shift is applied to 'n' and not
link_clock. For, link_clock=810000kHz, lane=1, n is 62E080h (810000*1*8) and right-shifting this by 1 does bring it under 80000h. Can you please clarify this?
Only *link* N is relevant here, not *data* N. The link M/N = pixel_clock/link_clock, and bpp and lane count etc. are irrelevant.
BR,
Jani.
Quoted 62 lines
The patch itself looks right and is well within the Spec.
Reviewed-by Dhinakaran Pandiyan <[email protected]>Quoted 56 lines
The pixel clock for modes that require a link clock >= 540000 kHz would also have several least significant bits zero. Unless the user provides a mode with an odd pixel clock value, we can reduce the numbers to reach the goal, with no loss in precision.
The DP spec even mentions sources making choices that "allow for static and relatively small Mvid and Nvid values", thus reducing the link M/N regardless of the sink in question seems justified.
Everything here is based on the work and information gathered by Clint Taylor <[email protected]>. This is just an iteration to reduce the parameters regardless of lane count, link rate, or sink.
Reference: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93578Tested-by Mads <[email protected]>Tested-by PJ <[email protected]>Tested-by François Guerraz <[email protected]>Tested-by Lev Popov <[email protected]>Tested-by Igor Krivenko <[email protected]>Cc Clint Taylor <[email protected]>Cc Anusha Srivatsa <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Daniel Vetter <[email protected]>Signed-off-by Jani Nikula <[email protected]>---
This is cc: stable material, but due to the slight risk of regressions (there's always the risk, however small, when you change parameters that affect all sinks) I'd prefer letting this simmer for a while, and asking for an explicit stable backport afterwards.
--- drivers/gpu/drm/i915/intel_display.c | 9 +++++++++ 1 file changed, 9 insertions(+)drivers/gpu/drm/i915/intel_display.c +9
-0@@ -6337,6 +6337,15 @@ intel_reduce_m_n_ratio(uint32_t *num, uint32_t *den)static void compute_m_n (unsigned int m,unsigned int n,uint32_t *ret_m,uint32_t *ret_n) { +/* +* Reduce M/N as much as possible without loss in precision. Several DP +* dongles in particular seem to be fussy about too large M/N values. +*/ +while ((m &1 ) ==0 && (n &1 ) ==0 ) { + m >>=1 ; + n >>=1 ; + } + *ret_n = min_t(unsigned int, roundup_pow_of_two(n), DATA_LINK_N_MAX); *ret_m =div_u64 ((uint64_t ) m * *ret_n, n);intel_reduce_m_n_ratio (ret_m, ret_n);
Signature · 5 lines
Jani Nikula, Intel Open Source Technology Center _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 18 lines · Jani Nikula
Quoted 12 lines · Dhinakaran
Quoted 10 lines · Jani Nikula
Several major vendor USB-C->HDMI converters, in particular the DA200, fail to recover a 5.4 GHz 1 lane signal if the link N is greater than
0x80000.The link M and N depend on the pixel clock and link clock ratio. With current code link N exceeds
0x80000only when link clock >= 540000 kHz. Except for the eDP intermediate link clocks, at least the four least significant bits are always zero. Just one bit shift right would be enough to bring even the DP 1.4 810000 kHz link clock under0x80000link N.Quoted 4 lines
I don't understand this part, the right shift is applied to 'n' and not
link_clock. For, link_clock=810000kHz, lane=1, n is 62E080h (810000*1*8) and right-shifting this by 1 does bring it under 80000h. Can you please clarify this?
Quoted 7 lines
Only *link* N is relevant here, not *data* N. The link M/N = pixel_clock/link_clock, and bpp and lane count etc. are irrelevant.
BR,
Jani.
I must have been blind to not see that, thanks for clarifying. -DK
Quoted 54 lines
Quoted 54 lines
The patch itself looks right and is well within the Spec.
Reviewed-by Dhinakaran Pandiyan <[email protected]>Quoted 49 lines
The pixel clock for modes that require a link clock >= 540000 kHz would also have several least significant bits zero. Unless the user provides a mode with an odd pixel clock value, we can reduce the numbers to reach the goal, with no loss in precision.
The DP spec even mentions sources making choices that "allow for static and relatively small Mvid and Nvid values", thus reducing the link M/N regardless of the sink in question seems justified.
Everything here is based on the work and information gathered by Clint Taylor <[email protected]>. This is just an iteration to reduce the parameters regardless of lane count, link rate, or sink.
Reference: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93578Tested-by Mads <[email protected]>Tested-by PJ <[email protected]>Tested-by François Guerraz <[email protected]>Tested-by Lev Popov <[email protected]>Tested-by Igor Krivenko <[email protected]>Cc Clint Taylor <[email protected]>Cc Anusha Srivatsa <[email protected]>Cc Ville Syrjälä <[email protected]>Cc Daniel Vetter <[email protected]>Signed-off-by Jani Nikula <[email protected]>---
This is cc: stable material, but due to the slight risk of regressions (there's always the risk, however small, when you change parameters that affect all sinks) I'd prefer letting this simmer for a while, and asking for an explicit stable backport afterwards.
--- drivers/gpu/drm/i915/intel_display.c | 9 +++++++++ 1 file changed, 9 insertions(+)drivers/gpu/drm/i915/intel_display.c +5
-0@@ -6337,6 +6337,15 @@ intel_reduce_m_n_ratio(uint32_t *num, uint32_t *den)static void compute_m_n (unsigned int m,unsigned int n,uint32_t *ret_m,uint32_t *ret_n) { +/* +* Reduce M/N as much as possible without loss in precision. Several DP +* dongles in particular seem to be fussy about too large M/N values. +*/ +while ((m &1 ) ==0 && (n &1 ) ==0 ) {
Quoted 9 lines
Quoted 8 lines
drivers/gpu/drm/i915/intel_display.c intel_reduce_m_n_ratio(uint32_t *num, uint32_t *den) · line 6351
Quoted 7 lines
drivers/gpu/drm/i915/intel_display.c +4
-0+ m >>=1 ; + n >>=1 ; + } + *ret_n = min_t(unsigned int, roundup_pow_of_two(n), DATA_LINK_N_MAX); *ret_m =div_u64 ((uint64_t ) m * *ret_n, n);intel_reduce_m_n_ratio (ret_m, ret_n);
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx