换回 5.10 好像 panfrost 开不了了。
是应该把 mesa 换回 maliG610 PPA的版本么？

update:已经降级并开启 panfrost

经过更新mesa终于可以进系统了。
顺便一提这个firmware的路径不是 /lib/firmware，放在那里还是会被找不到。

看日志发现是应该放在： /lib/firmware/arm/mali/arch10.8/ 下面
简单测试了一下，首先是firefox 开 GPU 之后页面不会乱撕裂了，浏览器好像也可以直接播放视频不会卡死，虽然不清楚是不是用了硬件加速。

接着就是发现没声音，和这个问题还不太一样：
https://cool-pi.com/topic/578/es8316-sound-issues-on-armbian-with-6-1-rockchip-kernel/2

就是单纯的没声音，麦克风和输出还有耳机和喇叭都没声音。

剩下的就是日志里一堆 rknpu 的报错，可能是 npu 模块还没编译？

休眠工作正常，开盖自动唤醒还是没有。

刚看到github更新了，立即clone 下来开始编译，结果新内核启动后黑屏了。

具体表现就是启动会显示 logo，但是会直接黑屏，不出现 gdm，按下关机键会回到命令行看到关机停止服务的各种指令。

我应该是把新的 firmware 放到目录了，但是感觉像是你说的没 firmware 的黑屏。

今天很晚了，明天再折腾研究一下。

@george 我懂，不过经过研究，我发现在键帽和导电胶之间垫上一层贴纸就可以极大程度的改善手感和按压不到底导致的吞字问题。

测试一阵后再来分享一下改造方式。

这个键盘手感差的原因就是导电胶和键帽之间的弹性有点差。需要每个按键都很大力的按下去才能精准触发。

@george 昨天已经看到了，辛苦
panthor的驱动是需要 mesa 编译实现吗，我看核心代码里已经有了。

还是不得不感慨，这个机器的键盘实在是太烂了。
经常有吞键，还有误触的情况。虽然感觉不太可能有机会改善了，但还是要吐槽一下。

最后我自己弄了个 22.04 的 USB 启动盘，在那个系统下编译通过了。

不过启动后加载这个 firmware 的确可以进系统，但是问题有点多。mesa 似乎也得更新对应版本，不然 panfrost 打不开，wayland 下 gtk 4 的东西都无法启动。

能力有限，还是等你们官方放出测试好的内核吧。

@george x86环境可以编译吗？我可以弄个wsl的ubuntu 22.04 试试

反复测试了几次，发现是RTW8852 网卡的模块编译不通过，难不成和我换网卡有关系么？

make[2]: *** [scripts/Makefile.build:500: drivers/media] Error 2
make[1]: *** [scripts/Makefile.build:500: drivers] Error 2
make: *** [Makefile:2012: .] Error 2
  CALL    scripts/checksyscalls.sh
  CC      drivers/media/platform/rockchip/cif/capture.o
  CC [M]  drivers/net/wireless/rockchip_wlan/rtw8852be/core/rtw_vht.o
drivers/media/platform/rockchip/cif/capture.c:4929:6: error: conflicting types for ‘rkcif_do_stop_stream’ due to enum/integer mismatch; have ‘void(struct rkcif_stream *, unsigned int)’ [-Werror=enum-int-mismatch]
 4929 | void rkcif_do_stop_stream(struct rkcif_stream *stream,
      |      ^~~~~~~~~~~~~~~~~~~~
In file included from drivers/media/platform/rockchip/cif/capture.c:24:
drivers/media/platform/rockchip/cif/dev.h:934:6: note: previous declaration of ‘rkcif_do_stop_stream’ with type ‘void(struct rkcif_stream *, enum rkcif_stream_mode)’
  934 | void rkcif_do_stop_stream(struct rkcif_stream *stream,
      |      ^~~~~~~~~~~~~~~~~~~~
drivers/media/platform/rockchip/cif/capture.c:6090:5: error: conflicting types for ‘rkcif_do_start_stream’ due to enum/integer mismatch; have ‘int(struct rkcif_stream *, unsigned int)’ [-Werror=enum-int-mismatch]
 6090 | int rkcif_do_start_stream(struct rkcif_stream *stream, unsigned int mode)
      |     ^~~~~~~~~~~~~~~~~~~~~
drivers/media/platform/rockchip/cif/dev.h:932:5: note: previous declaration of ‘rkcif_do_start_stream’ with type ‘int(struct rkcif_stream *, enum rkcif_stream_mode)’
  932 | int rkcif_do_start_stream(struct rkcif_stream *stream,
      |     ^~~~~~~~~~~~~~~~~~~~~
  CC      drivers/clocksource/timer-of.o
  CC      drivers/clocksource/timer-probe.o
drivers/net/wireless/rockchip_wlan/rtw8852be/core/rtw_vht.c: In function ‘rtw_reattach_vht_ies’:
drivers/net/wireless/rockchip_wlan/rtw8852be/core/rtw_vht.c:1419:27: error: the comparison will always evaluate as ‘true’ for the address of ‘IEs’ will never be NULL [-Werror=address]
 1419 |         if (pnetwork->IEs != NULL) {
      |                           ^~
In file included from drivers/net/wireless/rockchip_wlan/rtw8852be/include/drv_types.h:27,
                 from drivers/net/wireless/rockchip_wlan/rtw8852be/core/rtw_vht.c:17:
drivers/net/wireless/rockchip_wlan/rtw8852be/include/wlan_bssdef.h:249:13: note: ‘IEs’ declared here
  249 |         u8  IEs[MAX_IE_SZ];     /* (timestamp, beacon interval, and capability information) */
      |             ^~~
cc1: all warnings being treated as errors
make[6]: *** [scripts/Makefile.build:250: drivers/net/wireless/rockchip_wlan/rtw8852be/core/rtw_vht.o] Error 1
make[5]: *** [scripts/Makefile.build:500: drivers/net/wireless/rockchip_wlan/rtw8852be] Error 2
make[4]: *** [scripts/Makefile.build:500: drivers/net/wireless/rockchip_wlan] Error 2
make[3]: *** [scripts/Makefile.build:500: drivers/net/wireless] Error 2
make[2]: *** [scripts/Makefile.build:500: drivers/net] Error 2
make[2]: *** Waiting for unfinished jobs....
  CC      drivers/clocksource/mmio.o
  CC      drivers/clocksource/timer-rockchip.o
  CC      drivers/clocksource/arm_arch_timer.o
  CC      drivers/clocksource/dummy_timer.o
  CC      drivers/media/platform/rockchip/hdmirx/rk_hdmirx_class.o
  CC      drivers/media/platform/rockchip/hdmirx/rk_hdmirx.o
  CC      drivers/media/platform/rockchip/hdmirx/rk_hdmirx_cec.o
  AR      drivers/clocksource/built-in.a
  CC      drivers/media/platform/rockchip/hdmirx/rk_hdmirx_hdcp.o
cc1: all warnings being treated as errors
make[6]: *** [scripts/Makefile.build:250: drivers/media/platform/rockchip/cif/capture.o] Error 1
make[5]: *** [scripts/Makefile.build:500: drivers/media/platform/rockchip/cif] Error 2
make[5]: *** Waiting for unfinished jobs....
  AR      drivers/media/platform/rockchip/hdmirx/built-in.a
make[4]: *** [scripts/Makefile.build:500: drivers/media/platform/rockchip] Error 2
make[3]: *** [scripts/Makefile.build:500: drivers/media/platform] Error 2
make[3]: *** Waiting for unfinished jobs....
make[2]: *** [scripts/Makefile.build:500: drivers/media] Error 2
make[1]: *** [scripts/Makefile.build:500: drivers] Error 2
make[1]: *** Waiting for unfinished jobs....
make: *** [Makefile:2012: .] Error 2
cp: cannot stat 'arch/arm64/boot/Image.gz': No such file or directory
cp: cannot stat 'arch/arm64/boot/Image': No such file or directory
sed: can't read modules.order: No such file or directory
make: *** [Makefile:1576: __modinst_pre] Error 2
cp: cannot stat 'vmlinuz': No such file or directory
cp: cannot stat 'Image': No such file or directory

话说我也有兴趣尝试主动散热，因为现在被动散热对于一些稍微大一点的应用的确顶不住。

主动散热我看v10版本就有配件，这个可以单独提供吗？我找到了同模具 gemibook x pro 的风扇的，但是找不到卖铜管的。

还有好奇的就是，主动散热风扇的话支持变频吗，还是永远最大转速。

just use flatpak

https://flathub.org/apps/com.tencent.WeChat

there are many app on flathub with arch64 support.

Tried to compile kernel 6.1 but fail.

make[4]: *** [scripts/Makefile.build:500: drivers/media/platform/rockchip] Error 2
make[3]: *** [scripts/Makefile.build:500: drivers/media/platform] Error 2
make[2]: *** [scripts/Makefile.build:500: drivers/media] Error 2
make[1]: *** [scripts/Makefile.build:500: drivers] Error 2
make: *** [Makefile:2012: .] Error 2
cp: cannot stat 'arch/arm64/boot/Image.gz': No such file or directory
cp: cannot stat 'arch/arm64/boot/Image': No such file or directory
sed: can't read modules.order: No such file or directory
make: *** [Makefile:1576: __modinst_pre] Error 2
cp: cannot stat 'vmlinuz': No such file or directory
cp: cannot stat 'Image': No such file or directory

and

CC      drivers/media/platform/rockchip/ispp/ispp.o
AR      drivers/media/firewire/built-in.a
AR      drivers/media/platform/rockchip/rkisp1/built-in.a
AR      drivers/media/spi/built-in.a
CC [M]  drivers/media/spi/cxd2880-spi.o
cc1: all warnings being treated as errors
make[6]: *** [scripts/Makefile.build:250: drivers/media/platform/rockchip/cif/capture.o] Error 1
make[5]: *** [scripts/Makefile.build:500: drivers/media/platform/rockchip/cif] Error 2
make[5]: *** Waiting for unfinished jobs....
AR      drivers/media/test-drivers/built-in.a
AR      drivers/media/radio/built-in.a

CC [M]  drivers/media/tuners/tda8290.o
CC [M]  drivers/media/tuners/tda9887.o
CC      drivers/media/common/videobuf2/videobuf2-dma-contig.o
CC      drivers/media/v4l2-core/v4l2-i2c.o
drivers/media/platform/rockchip/cif/capture.c:4929:6: error: conflicting types for ‘rkcif_do_stop_stream’ due to enum/integer mismatch; have ‘void(struct rkcif_stream *, unsigned int)’ [-Werror=enum-int-mismatch]
4929 | void rkcif_do_stop_stream(struct rkcif_stream *stream,
    |      ^~~~~~~~~~~~~~~~~~~~
In file included from drivers/media/platform/rockchip/cif/capture.c:24:
drivers/media/platform/rockchip/cif/dev.h:934:6: note: previous declaration of ‘rkcif_do_stop_stream’ with type ‘void(struct rkcif_stream *, enum rkcif_stream_mode)’
934 | void rkcif_do_stop_stream(struct rkcif_stream *stream,
    |      ^~~~~~~~~~~~~~~~~~~~
drivers/media/platform/rockchip/cif/capture.c:6090:5: error: conflicting types for ‘rkcif_do_start_stream’ due to enum/integer mismatch; have ‘int(struct rkcif_stream *, unsigned int)’ [-Werror=enum-int-mismatch]
6090 | int rkcif_do_start_stream(struct rkcif_stream *stream, unsigned int mode)
    |     ^~~~~~~~~~~~~~~~~~~~~
drivers/media/platform/rockchip/cif/dev.h:932:5: note: previous declaration of ‘rkcif_do_start_stream’ with type ‘int(struct rkcif_stream *, enum rkcif_stream_mode)’
932 | int rkcif_do_start_stream(struct rkcif_stream *stream,
    |     ^~~~~~~~~~~~~~~~~~~~~
CC [M]  drivers/media/dvb-frontends/l64781.o
CC [M]  drivers/media/dvb-frontends/lg2160.o
CC [M]  drivers/media/tuners/tea5761.o

@george Many thanks for sharing.

I will try it with kernel 6.1

@yusuf-tmp Thanks for your sharing.

How about kernel 6.1 performance on cool pi notebook?

WIth kernel 5.10 right now it's annoying about no gpu acceleration on firefox and high power cosuming.

Here is more about schedutil

https://github.com/ThomasKaiser/Knowledge/blob/master/articles/schedutil-and-EAS.md

Seems ondemand is better choice.

I just found something useful about improve cool pi laptop perfomance and efficiency.

https://github.com/ThomasKaiser/Knowledge/blob/master/articles/Quick_Preview_of_ROCK_5B.md#important-insights-and-suggested-optimisations

I tried and it works quite well.

Just compile kernel with CONFIG_CPU_FREQ_DEFAULT_GOV_ONDEMAND=y and use a conf file.

@george 期待下一代产品，会保持关注的。
cool pi 的技术支持是我见过最专业和迅速的，有问必答。

@george 2K好屏幕有很多，但是如果是预装 Ubuntu 的话我不太建议用 2K，因为 Gnome 的 “非整数缩放”做的很烂，只有 KDE 会好一些。

升级 2K 屏的话最好就得把 DE 换成 KDE 开 150% 了，Gnome 的话会很难受。

我是网络运维，平时有兼职写稿子，这个本子用起来的确不错，除了键盘敲起来和我原来的 ThinkPad 比起来还是有差距，不过这也没办法。

顺便一提我看到你们在 hackaday 的链接了：
https://hackaday.io/project/196668-genbook-rk3588/log/230987-instructions-of-genbook-rk3588

里面列出了 lcd 升级的表格，我升级的是NV140FHM-N66 这块面板，比起列表里那个 N61 来说，亮度低了一点，但是是 DC 调光，比较护眼。目前使用完美兼容。

本来想升级 4K 的，但是怕功耗顶不住，暂时搁置了，面板也选好了，大概率是可以用的。 Mali G610 应该是可以 hold 住 4K 屏幕的。

@george
还有就是很好奇，3588这个soc会有类似“温度墙”的东西吗？
昨天编译内核发现温度已经飙到83摄氏度了，会影响性能吗？