linux: update to 6.4.y #7837

heitbaum · 2023-05-13T06:45:12Z

6.4 - Linux Kernel release
6.4.1 - 31 patches

errors / fixes / issues / regressions / todo

incorporate Wireless Intel AX101
- https://www.spinics.net/lists/linux-wireless/msg236143.html
nvidia update to 470.182.03 and patch
rtw88: drop upstreamed patches included in kernel 6.4
TODO: Add working 6.3 rtw88 patch to Amlogic
TODO: Rockchip
- review WIP patches
update to Linux 6.4.1
update iwlwifi firmware for 6.4 - core78

log

6.4.1 Build tested on all of:

PROJECT=Allwinner ARCH=arm DEVICE=A64 s/build linux
PROJECT=Allwinner ARCH=arm DEVICE=H3 s/build linux
PROJECT=Allwinner ARCH=arm DEVICE=H5 s/build linux
PROJECT=Allwinner ARCH=arm DEVICE=H6 s/build linux
PROJECT=Rockchip ARCH=arm DEVICE=RK3288 s/build linux
PROJECT=Rockchip ARCH=arm DEVICE=RK3328 s/build linux
PROJECT=Rockchip ARCH=arm DEVICE=RK3399 s/build linux
PROJECT=NXP ARCH=arm DEVICE=iMX6 s/build linux
PROJECT=NXP ARCH=arm DEVICE=iMX8 s/build linux
PROJECT=Qualcomm ARCH=arm DEVICE=Dragonboard s/build linux - u-boot build failure
PROJECT=Generic ARCH=x86_64 DEVICE=Generic s/build linux
PROJECT=Generic ARCH=x86_64 DEVICE=Generic-legacy s/build linux
PROJECT=Samsung ARCH=arm DEVICE=Exynos s/build linux

6.3.y Run tested on all of:

Allwinner all - tested - TBA - jernejsk
Allwinner H6 (Tanix TX6) - 6.4.1 - heitbaum
Generic Generic (Intel ADL - NUC12) - 6.4.1 - heitbaum
NXP iMX6 (Cubox-i4Pro) - TBA - heitbaum
Rockchip RK3399pro (Rock Pi N10) - TBA - heitbaum
RK all - tested - TBA - knaerzche
Samsung Exynos (Hardkernel ODROID XU4) - TBA - heitbaum

——

Fixed: ~~Issues on Tanix TX6 - 8822bs/cs - fixed with byte-wise-word-io patch~~

8822bs
[    8.817489] rtw_8822bs mmc1:0001:1: Firmware version 27.2.0, H2C version 13
[    8.817786] sunxi-mmc 4021000.mmc: unaligned scatterlist: os f80 length 2
[    8.817796] sunxi-mmc 4021000.mmc: map DMA failed
[    8.817804] rtw_8822bs mmc1:0001:1: sdio read16 failed (0x10230): -22
...
[    8.818259] rtw_8822bs mmc1:0001:1: failed to download rsvd page
[    8.818313] rtw_8822bs mmc1:0001:1: failed to download firmware
...
[    8.819096] rtw_8822bs mmc1:0001:1: failed to setup chip efuse info
[    8.819103] rtw_8822bs mmc1:0001:1: failed to setup chip information
[    8.819323] rtw_8822bs: probe of mmc1:0001:1 failed with error -16

8822cs
[    7.426639] rtw_8822cs mmc1:0001:1: WOW Firmware version 9.9.4, H2C version 15
[    7.428342] rtw_8822cs mmc1:0001:1: Firmware version 9.9.15, H2C version 15
[    7.437851] sunxi-mmc 4021000.mmc: unaligned scatterlist: os f80 length 2
[    7.437891] sunxi-mmc 4021000.mmc: map DMA failed
[    7.437901] rtw_8822cs mmc1:0001:1: sdio read16 failed (0x10230): -22
...
[    7.438363] rtw_8822cs mmc1:0001:1: sdio write16 failed (0x10204): -22
[    7.438393] rtw_8822cs mmc1:0001:1: failed to download rsvd page
[    7.438444] rtw_8822cs mmc1:0001:1: failed to download firmware
...
[    7.438928] rtw_8822cs mmc1:0001:1: failed to setup chip efuse info
[    7.438944] rtw_8822cs mmc1:0001:1: failed to setup chip information
[    7.439218] rtw_8822cs: probe of mmc1:0001:1 failed with error -16

jernejsk · 2023-05-13T07:15:13Z

See https://lore.kernel.org/linux-wireless/880bb4ed-f3a5-2d7f-db10-fec65087dd05@lwfinger.net/T/#mf53120e175d8015c12a8ef57d2643da2ac34a74f for wifi issue

heitbaum · 2023-05-13T08:56:23Z

See https://lore.kernel.org/linux-wireless/880bb4ed-f3a5-2d7f-db10-fec65087dd05@lwfinger.net/T/#mf53120e175d8015c12a8ef57d2643da2ac34a74f for wifi issue

I can confirm that both TX6 (822bs and 8822cs have running WiFi and Bluetooth with 6.4-rc1 and the byte-wise-word-io patch. I havent done any performance tests / ... But connectivity was no issue.

With HCI_QUIRK_BROKEN_LOCAL_EXT_FEATURES_PAGE_2 now in 6.4-rc1 - I'll tidy up the BT 8822bs patch and submit.

heitbaum · 2023-05-14T03:37:31Z

iperf testing has no dmesg errors - tested with both ipv4/ipv6 and both arm/aarch64 user lands

tx6-8822cs:~ # iperf --client nuc12.local -t 60
...
[  5]  57.00-58.00  sec  4.85 MBytes  40.7 Mbits/sec    0   1.06 MBytes
[  5]  58.00-59.00  sec  5.40 MBytes  45.3 Mbits/sec    0   1.06 MBytes
[  5]  59.00-60.00  sec  5.10 MBytes  42.7 Mbits/sec    0   1.06 MBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-60.00  sec   351 MBytes  49.1 Mbits/sec    0             sender
[  5]   0.00-60.04  sec   347 MBytes  48.5 Mbits/sec                  receiver

tx6-8822bs:~ # iperf --client nuc12.local -t 60
...
[  5]  57.00-58.00  sec  4.11 MBytes  34.5 Mbits/sec    0    650 KBytes
[  5]  58.00-59.00  sec  4.85 MBytes  40.7 Mbits/sec    0    650 KBytes
[  5]  59.00-60.00  sec  4.85 MBytes  40.7 Mbits/sec    0    650 KBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-60.00  sec   291 MBytes  40.7 Mbits/sec    0             sender
[  5]   0.00-60.03  sec   289 MBytes  40.4 Mbits/sec                  receiver


nuc12:~ # iperf --client tx6-8822cs -t 60
...
[  5]  57.00-58.00  sec  7.61 MBytes  63.9 Mbits/sec    0   3.15 MBytes
[  5]  58.00-59.00  sec  7.92 MBytes  66.4 Mbits/sec    0   3.15 MBytes
[  5]  59.00-60.00  sec  8.60 MBytes  72.1 Mbits/sec    0   3.15 MBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-60.00  sec   493 MBytes  68.9 Mbits/sec    0             sender
[  5]   0.00-60.00  sec   492 MBytes  68.8 Mbits/sec                  receiver

nuc12:~ # iperf --client tx6-8822bs -t 60
...
[  5]  57.00-58.00  sec  4.79 MBytes  40.2 Mbits/sec    0   3.15 MBytes
[  5]  58.00-59.00  sec  4.30 MBytes  36.1 Mbits/sec    0   3.15 MBytes
[  5]  59.00-60.00  sec  4.11 MBytes  34.5 Mbits/sec    0   3.15 MBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-60.00  sec   260 MBytes  36.3 Mbits/sec    0             sender
[  5]   0.00-60.00  sec   259 MBytes  36.3 Mbits/sec                  receiver

The Allwinner sunxi-mmc controller cannot handle word (16 bit) transfers. So and sdio_{read,write}w fails with messages like the following example using an RTL8822BS (but the same problems were also observed with RTL8822CS and RTL8723DS chips): rtw_8822bs mmc1:0001:1: Firmware version 27.2.0, H2C version 13 sunxi-mmc 4021000.mmc: unaligned scatterlist: os f80 length 2 sunxi-mmc 4021000.mmc: map DMA failed rtw_8822bs mmc1:0001:1: sdio read16 failed (0x10230): -22 Use two consecutive single byte accesses for word operations instead. It turns out that upon closer inspection this is also what the vendor driver does, even though it does have support for sdio_{read,write}w. So we can conclude that the rtw88 chips do support word access but only on SDIO controllers that also support it. Since there's no way to detect if the controller supports word access or not the rtw88 sdio driver switches to the easiest approach: avoiding word access. Reported-by: Larry Finger <Larry.Finger@lwfinger.net> Closes: https://lore.kernel.org/linux-wireless/527585e5-9cdd-66ed-c3af-6da162f4b720@lwfinger.net/ Reported-by: Rudi Heitbaum <rudi@heitbaum.com> Link: LibreELEC/LibreELEC.tv#7837 (comment) Fixes: 65371a3 ("wifi: rtw88: sdio: Add HCI implementation for SDIO based chipsets") Reviewed-by: Ping-Ke Shih <pkshih@realtek.com> Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>

heitbaum · 2023-06-24T04:41:27Z

@Kwiboo / @knaerzche - are these rebases ok? Otherwise let me know. If so I squash (change from wip) and finalise 6.4.

3ccfdf5653bee1b575fcd39ddafb5782379f4b35
3a8937ba828b809c56e95765fe4d7f2ed5e926aa

heitbaum · 2023-06-25T04:45:59Z

@Kwiboo / @knaerzche - are these rebases ok? Otherwise let me know. If so I squash (change from wip) and finalise 6.4.

3ccfdf5

3a8937b

I can take a closer look, the dropped patches should probably be reworked and not just dropped.

Currently slowly working on upstreaming rockchip hdmi 2.0 patches and will rework as progress is made.

@Kwiboo I’ve gone through the code today, the 3 dropped patches were across the same function dw_hdmi_rockchip_mode_valid. Diffs are below, few differences in that MODE_CLOCK_HIGH, ycbcr_420_allowed and no specific handling for 340000. I can see where the code has diverged but come back together in the mpll_cfg, but not quite sure on rebasing it all back together.

the brute force revert 6.5 to 6.4-LE patch is here 40ff3634f71b8bf1769d05a565e7275b6ecbac51 which is easier to read

upstream change 6.4 to 6.5

@@ -241,23 +250,39 @@
 }
 
 static enum drm_mode_status
-dw_hdmi_rockchip_mode_valid(struct dw_hdmi *hdmi, void *data,
+dw_hdmi_rockchip_mode_valid(struct dw_hdmi *dw_hdmi, void *data,
                            const struct drm_display_info *info,
                            const struct drm_display_mode *mode)
 {
+       struct rockchip_hdmi *hdmi = data;
        const struct dw_hdmi_mpll_config *mpll_cfg = rockchip_mpll_cfg;
        int pclk = mode->clock * 1000;
-       bool valid = false;
+       bool exact_match = hdmi->plat_data->phy_force_vendor;
        int i;
 
+       if (hdmi->ref_clk) {
+               int rpclk = clk_round_rate(hdmi->ref_clk, pclk);
+
+               if (abs(rpclk - pclk) > pclk / 1000)
+                       return MODE_NOCLOCK;
+       }
+
        for (i = 0; mpll_cfg[i].mpixelclock != (~0UL); i++) {
-               if (pclk == mpll_cfg[i].mpixelclock) {
-                       valid = true;
-                       break;
-               }
+               /*
+                * For vendor specific phys force an exact match of the pixelclock
+                * to preserve the original behaviour of the driver.
+                */
+               if (exact_match && pclk == mpll_cfg[i].mpixelclock)
+                       return MODE_OK;
+               /*
+                * The Synopsys phy can work with pixelclocks up to the value given
+                * in the corresponding mpll_cfg entry.
+                */
+               if (!exact_match && pclk <= mpll_cfg[i].mpixelclock)
+                       return MODE_OK;
        }
 
-       return (valid) ? MODE_OK : MODE_BAD;
+       return MODE_BAD;
 }
 
 static void dw_hdmi_rockchip_encoder_disable(struct drm_encoder *encoder)

Downstream LE patches 6.4 to 6.4-LE

@@ -245,19 +245,31 @@
                            const struct drm_display_info *info,
                            const struct drm_display_mode *mode)
 {
-       const struct dw_hdmi_mpll_config *mpll_cfg = rockchip_mpll_cfg;
-       int pclk = mode->clock * 1000;
-       bool valid = false;
-       int i;
-
-       for (i = 0; mpll_cfg[i].mpixelclock != (~0UL); i++) {
-               if (pclk == mpll_cfg[i].mpixelclock) {
-                       valid = true;
-                       break;
-               }
+       struct dw_hdmi_plat_data *pdata = (struct dw_hdmi_plat_data *)data;
+       const struct dw_hdmi_mpll_config *mpll_cfg = pdata->mpll_cfg;
+       int clock = mode->clock;
+       unsigned int i = 0;
+
+       if (pdata->ycbcr_420_allowed && drm_mode_is_420(info, mode) &&
+           (info->color_formats & DRM_COLOR_FORMAT_YCBCR420)) {
+               clock /= 2;
+               mpll_cfg = pdata->mpll_cfg_420;
+       }
+
+       if ((!mpll_cfg && clock > 340000) ||
+           (info->max_tmds_clock && clock > info->max_tmds_clock))
+               return MODE_CLOCK_HIGH;
+
+       if (mpll_cfg) {
+               while ((clock * 1000) < mpll_cfg[i].mpixelclock &&
+                      mpll_cfg[i].mpixelclock != (~0UL))
+                      i++;
+
+               if (mpll_cfg[i].mpixelclock == (~0UL))
+                       return MODE_CLOCK_HIGH;
        }
 
-       return (valid) ? MODE_OK : MODE_BAD;
+       return MODE_OK;
 }
 
 static void dw_hdmi_rockchip_encoder_disable(struct drm_encoder *encoder)

Resulting difference between 6.4-LE and 6.5


@@ -241,35 +250,39 @@
 }
 
 static enum drm_mode_status
-dw_hdmi_rockchip_mode_valid(struct dw_hdmi *hdmi, void *data,
+dw_hdmi_rockchip_mode_valid(struct dw_hdmi *dw_hdmi, void *data,
                            const struct drm_display_info *info,
                            const struct drm_display_mode *mode)
 {
-       struct dw_hdmi_plat_data *pdata = (struct dw_hdmi_plat_data *)data;
-       const struct dw_hdmi_mpll_config *mpll_cfg = pdata->mpll_cfg;
-       int clock = mode->clock;
-       unsigned int i = 0;
-
-       if (pdata->ycbcr_420_allowed && drm_mode_is_420(info, mode) &&
-           (info->color_formats & DRM_COLOR_FORMAT_YCBCR420)) {
-               clock /= 2;
-               mpll_cfg = pdata->mpll_cfg_420;
-       }
-
-       if ((!mpll_cfg && clock > 340000) ||
-           (info->max_tmds_clock && clock > info->max_tmds_clock))
-               return MODE_CLOCK_HIGH;
-
-       if (mpll_cfg) {
-               while ((clock * 1000) < mpll_cfg[i].mpixelclock &&
-                      mpll_cfg[i].mpixelclock != (~0UL))
-                      i++;
-
-               if (mpll_cfg[i].mpixelclock == (~0UL))
-                       return MODE_CLOCK_HIGH;
+       struct rockchip_hdmi *hdmi = data;
+       const struct dw_hdmi_mpll_config *mpll_cfg = rockchip_mpll_cfg;
+       int pclk = mode->clock * 1000;
+       bool exact_match = hdmi->plat_data->phy_force_vendor;
+       int i;
+
+       if (hdmi->ref_clk) {
+               int rpclk = clk_round_rate(hdmi->ref_clk, pclk);
+
+               if (abs(rpclk - pclk) > pclk / 1000)
+                       return MODE_NOCLOCK;
+       }
+
+       for (i = 0; mpll_cfg[i].mpixelclock != (~0UL); i++) {
+               /*
+                * For vendor specific phys force an exact match of the pixelclock
+                * to preserve the original behaviour of the driver.
+                */
+               if (exact_match && pclk == mpll_cfg[i].mpixelclock)
+                       return MODE_OK;
+               /*
+                * The Synopsys phy can work with pixelclocks up to the value given
+                * in the corresponding mpll_cfg entry.
+                */
+               if (!exact_match && pclk <= mpll_cfg[i].mpixelclock)
+                       return MODE_OK;
        }
 
-       return MODE_OK;
+       return MODE_BAD;
 }
 
 static void dw_hdmi_rockchip_encoder_disable(struct drm_encoder *encoder)

…6.4-rc1 in projects/Rockchip/patches/linux/default/linux-1000-drm-rockchip.patch need to review drivers/gpu/drm/rockchip/dw_hdmi-rockchip.c in line with - https://github.com/torvalds/linux/commits/master/drivers/gpu/drm/rockchip/dw_hdmi-rockchip.c - torvalds/linux@d13b10e - torvalds/linux@de13db3 Drop these patches - later patch to reinstate downstream changes vop_crtc_mode_valid updated.

jira LE-1907 Rebuild_History Non-Buildable kernel-4.18.0-512.el8 commit-author Martin Blumenstingl <martin.blumenstingl@googlemail.com> commit cb0ddaa The Allwinner sunxi-mmc controller cannot handle word (16 bit) transfers. So and sdio_{read,write}w fails with messages like the following example using an RTL8822BS (but the same problems were also observed with RTL8822CS and RTL8723DS chips): rtw_8822bs mmc1:0001:1: Firmware version 27.2.0, H2C version 13 sunxi-mmc 4021000.mmc: unaligned scatterlist: os f80 length 2 sunxi-mmc 4021000.mmc: map DMA failed rtw_8822bs mmc1:0001:1: sdio read16 failed (0x10230): -22 Use two consecutive single byte accesses for word operations instead. It turns out that upon closer inspection this is also what the vendor driver does, even though it does have support for sdio_{read,write}w. So we can conclude that the rtw88 chips do support word access but only on SDIO controllers that also support it. Since there's no way to detect if the controller supports word access or not the rtw88 sdio driver switches to the easiest approach: avoiding word access. Reported-by: Larry Finger <Larry.Finger@lwfinger.net> Closes: https://lore.kernel.org/linux-wireless/527585e5-9cdd-66ed-c3af-6da162f4b720@lwfinger.net/ Reported-by: Rudi Heitbaum <rudi@heitbaum.com> Link: LibreELEC/LibreELEC.tv#7837 (comment) Fixes: 65371a3 ("wifi: rtw88: sdio: Add HCI implementation for SDIO based chipsets") Reviewed-by: Ping-Ke Shih <pkshih@realtek.com> Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com> Signed-off-by: Kalle Valo <kvalo@kernel.org> Link: https://lore.kernel.org/r/20230515195043.572375-1-martin.blumenstingl@googlemail.com (cherry picked from commit cb0ddaa) Signed-off-by: Jonathan Maple <jmaple@ciq.com>

heitbaum added WORK IN PROGRESS UPDATE MERGE BLOCKED PACKAGE KERNEL ALLWINNER ROCKCHIP NXP SAMSUNG GENERIC LE 12.0 labels May 13, 2023

heitbaum force-pushed the kernel64 branch from 112ebe5 to 9bca0ff Compare May 13, 2023 08:49

heitbaum force-pushed the kernel64 branch 5 times, most recently from 0a7adbe to 1d67891 Compare May 14, 2023 22:30

heitbaum force-pushed the kernel64 branch 7 times, most recently from ea3141e to 566dbd8 Compare June 24, 2023 04:37

heitbaum force-pushed the kernel64 branch from 40ff363 to 467b38d Compare June 26, 2023 06:15

heitbaum marked this pull request as ready for review June 26, 2023 06:18

heitbaum added 18 commits June 26, 2023 19:24

iwlwifi-firmware: update to 03bf32c

08c3799

linux: rtw88: drop upstreamed patches included in kernel 6.4

6e9e25e

linux (NXP iMX8): rebase patches for linux 6.4-rc1

3547955

linux (Rockchip): rebase patches for linux 6.4-rc1

3d3206a

linux (Allwinner aarch64): update .config for 6.4

45e86c9

linux (Allwinner arm): update .config for 6.4

2267961

linux (Generic): update .config for 6.4

952e45f

linux (NXP iMX6): update .config for 6.4

4bd8bec

linux (NXP iMX8): update .config for 6.4

ba0d757

linux (Qualcomm): update .config for 6.4

fd23474

linux (Rockchip RK3288): update .config for 6.4

5728a6e

linux (Rockchip RK3328): update .config for 6.4

3498c6a

linux (Rockchip RK3399): update .config for 6.4

d3c3026

linux (Samsung): update .config for 6.4

9aabdaf

xf86-video-nvidia: update to 470.182.03

9adcead

linux (Rockchip): patch for dw_hdmi-rockchip to revert 6.4 to 6.3-LE

b87714a

linux: update to 6.4

b41e3ba

heitbaum force-pushed the kernel64 branch from 467b38d to b41e3ba Compare June 26, 2023 19:40

heitbaum removed WORK IN PROGRESS MERGE BLOCKED labels Jun 26, 2023

linux: update to 6.4.1

b208d29

CvH approved these changes Jul 4, 2023

View reviewed changes

CvH merged commit db76177 into LibreELEC:master Jul 4, 2023

heitbaum deleted the kernel64 branch July 4, 2023 21:57

heitbaum mentioned this pull request Dec 22, 2023

Since #7837 kernel 6.4.y Rock64 does not display on a 4K screen #8446

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

linux: update to 6.4.y #7837

linux: update to 6.4.y #7837

heitbaum commented May 13, 2023 •

edited

Loading

jernejsk commented May 13, 2023

heitbaum commented May 13, 2023

heitbaum commented May 14, 2023

heitbaum commented Jun 24, 2023

heitbaum commented Jun 25, 2023

linux: update to 6.4.y #7837

linux: update to 6.4.y #7837

Conversation

heitbaum commented May 13, 2023 • edited Loading

log

6.4.1 Build tested on all of:

6.3.y Run tested on all of:

jernejsk commented May 13, 2023

heitbaum commented May 13, 2023

heitbaum commented May 14, 2023

heitbaum commented Jun 24, 2023

heitbaum commented Jun 25, 2023

heitbaum commented May 13, 2023 •

edited

Loading