Krzysztof Kozlowski
Posts
191Following
31Followers
191🇵🇱 🇪🇺 🇰🇷 🇮🇱 🇺🇦 🇨ðŸ‡
IRC: krzk
Kernel work related account. Other accounts of mine: @krzk@mastodon.social
Krzysztof Kozlowski
krzk
Edited 11 months ago
Tensor is coming to the next Linux kernel: v6.8!
I just applied last patches for Linux kernel bringing support for the Google Tensor GS101 SoC (derivative of Samsung Exynos):
https://lore.kernel.org/linux-samsung-soc/170249498744.308886.17508790332822828341.b4-ty@linaro.org/T/#u
Commits will hit tomorrow's linux-next and probably next week I will send pull request with them to SoC (@arnd@society.oftrolls.com and Olof) for the next merge window. Good job Peter Griffin and @LinaroLtd :)
I just applied last patches for Linux kernel bringing support for the Google Tensor GS101 SoC (derivative of Samsung Exynos):
https://lore.kernel.org/linux-samsung-soc/170249498744.308886.17508790332822828341.b4-ty@linaro.org/T/#u
Commits will hit tomorrow's linux-next and probably next week I will send pull request with them to SoC (@arnd@society.oftrolls.com and Olof) for the next merge window. Good job Peter Griffin and @LinaroLtd :)
3
13
39
Krzysztof Kozlowski
repeated
repeated
I can finally reveal some research I've been involved with over the past year or so.
We (@redford, @mrtick and I) have reverse engineered the PLC code of NEWAG Impuls EMUs. These trains were locking up for arbitrary reasons after being serviced at third-party workshops. The manufacturer argued that this was because of malpractice by these workshops, and that they should be serviced by them instead of third parti
es.
1/4
6
58
2
@kernellogger What was making me sad with such bisection, that after few hours of bisecting, not fully automated so with my personal (not work!) time spent on this, I send a report or even a fix and the maintainer squashes it with original commit. No credits about my work, no Reported-by, nothing for my few hours of personal time which I could spend on eating skittles.
After few of such cases, I stopped entirely reporting linux-next issues hit on my systems, unless the issue is visible for like two weeks. After two weeks, if my CI is still finding the issue, then there is a chance I will be credited with finding it...
After few of such cases, I stopped entirely reporting linux-next issues hit on my systems, unless the issue is visible for like two weeks. After two weeks, if my CI is still finding the issue, then there is a chance I will be credited with finding it...
1
0
3
Krzysztof Kozlowski
krzk
@kernellogger Bisecting should be on or start with linux-next. I don't think we should be sending fixes to Linus' somehow faster. Rather people should learn to work on linux-next.
1
0
0
Krzysztof Kozlowski
repeated
repeated
Oleksandr Natalenko, MSE
oleksandr@natalenko.nameA reminder about what Muscovy did to us in 1932/33: https://en.wikipedia.org/wiki/Holodomor. Today is the Remembrance Day: https://en.wikipedia.org/wiki/Holodomor_Memorial_Day
0
3
4
Krzysztof Kozlowski
repeated
repeated
Anne Applebaum
anneapplebaum@journa.hostToday is the annual day of remembrance for the Holodomor, the Ukrainian famine. 90 years ago Stalin sent activists to confiscate food from Ukrainian peasants. Millions died.
To mark this day, Putin sent dozens of drones to attack Kyiv.
Like Stalin, Putin wants to erase Ukraine.
2
8
2
Krzysztof Kozlowski
krzk
@cas @monsieuricon I cannot access BIOS on company laptop :) and even if I could I think it's still Lenovo's fault - you don't swap keys in a keyboard after 30 years of having the CTRL at that place :) (Y looks at Z.... anyone uses German keyboards? :) )
1
0
1
Krzysztof Kozlowski
krzk
@monsieuricon There is a special place in hell for designers who swapped CTRL and Fn... I am staying away from Lenovo, but it is tricky - many companies buy these for employees. :(
1
0
0
Krzysztof Kozlowski
krzk
Edited 11 months ago
Looking at Devicetree sources changes (DTS) could give some insight into which ARM64 platforms are the most active. DTS represents the real hardware, thus any new hardware-block for an existing SoC, new SoC or new board, require new DTS changes. It's an approximation, easy to measure and still quite informative. Let's take a look - the most developed ARM64 platforms in the upstream Linux kernel since last LTS (v6.1):
$ git diff --dirstat=changes,1 v6.1..v6.6 -- arch/arm64/boot/dts/ | sort -nr
44.6% arch/arm64/boot/dts/qcom/
12.2% arch/arm64/boot/dts/ti/
10.4% arch/arm64/boot/dts/freescale/
10.1% arch/arm64/boot/dts/rockchip/
5.0% arch/arm64/boot/dts/nvidia/
4.3% arch/arm64/boot/dts/mediatek/
3.9% arch/arm64/boot/dts/renesas/
3.5% arch/arm64/boot/dts/apple/
3.4% (the rest)
2.2% arch/arm64/boot/dts/amlogic/
Almost 45% of all changes to ARM64-related hardware in the upstream kernel were for Qualcomm SoCs and boards. That's quite a stunning number.
$ git diff --dirstat=changes,1 v6.1..v6.6 -- arch/arm64/boot/dts/ | sort -nr
44.6% arch/arm64/boot/dts/qcom/
12.2% arch/arm64/boot/dts/ti/
10.4% arch/arm64/boot/dts/freescale/
10.1% arch/arm64/boot/dts/rockchip/
5.0% arch/arm64/boot/dts/nvidia/
4.3% arch/arm64/boot/dts/mediatek/
3.9% arch/arm64/boot/dts/renesas/
3.5% arch/arm64/boot/dts/apple/
3.4% (the rest)
2.2% arch/arm64/boot/dts/amlogic/
Almost 45% of all changes to ARM64-related hardware in the upstream kernel were for Qualcomm SoCs and boards. That's quite a stunning number.
0
1
5
Krzysztof Kozlowski
repeated
repeated
K. Ryabitsev
monsieuricon
When patchwork integration is configured, b4 will now retrieve the CI status of each patch as well.
5
9
34
Krzysztof Kozlowski
krzk
@ptesarik @oleksandr @suse @kernellogger v2.6.40 is a development kernel, so it's actually perfectly fine and valid release. :)
1
0
0
@ptesarik @oleksandr @kernellogger @suse Appreciate @suse kernels? That's bollocks! Suse has the same 4baf12181 commit in their SLE15-SP6 tree! Whole team of engineers backported the same commit questioned here, instead of working on upstream stable backports, and this should be the argument to "appreciate" @suse.
You get absolutely nothing with enterprise kernels. If bug is in the mainline or upstream stable, Suse and Redhat have the bug as well.
You get absolutely nothing with enterprise kernels. If bug is in the mainline or upstream stable, Suse and Redhat have the bug as well.
0
0
0
@oleksandr @kernellogger @ptesarik @suse Although to be honest, that v2.6 (v2.6.32 to be specific) was kept as SLTS by the community till 2016.
0
0
0
Krzysztof Kozlowski
krzk
@oleksandr @suse @kernellogger @ptesarik
v4.14 is a SLTS, so it is old indeed, but we do not compare it with RHEL v4.18. Few years ago (2020) RHEL was still updating v3.10. This is the Uber-Franken-kernel we talk about.
In 2018 RHEL released v2.6 kernel. v2.6, can you imagine?
And do you know that also features get backported to enterprise distro Frankenkernels?
v4.14 is a SLTS, so it is old indeed, but we do not compare it with RHEL v4.18. Few years ago (2020) RHEL was still updating v3.10. This is the Uber-Franken-kernel we talk about.
In 2018 RHEL released v2.6 kernel. v2.6, can you imagine?
And do you know that also features get backported to enterprise distro Frankenkernels?
2
0
1
Krzysztof Kozlowski
krzk
@oleksandr @suse @kernellogger @ptesarik "Frankenkernel" is an very old kernel which consists of thousands of backports, thus like Frankenstein.
Stable kernels are not that, because they cease to exist at some point. You need to move to newer kernel, thus backporting stops. Beyond that point, it's the distro (Redhat, Suse, also Canonical for some extended support) who creates the Frankenkernel.
And if you ever want to call stable upstream a "Frankenkernel", then Suse and Redhat create Uber-super-Franken-master-kernel...
Stable kernels are not that, because they cease to exist at some point. You need to move to newer kernel, thus backporting stops. Beyond that point, it's the distro (Redhat, Suse, also Canonical for some extended support) who creates the Frankenkernel.
And if you ever want to call stable upstream a "Frankenkernel", then Suse and Redhat create Uber-super-Franken-master-kernel...
1
0
2
0
0
2
@ptesarik @oleksandr @kernellogger @suse Oh yes, the franken-kernels with selected fixes for 4 year old kernel... Or maybe even older, because Suse customers do not like to update and test their stuff (reasonable, no one likes testing!).
1
0
2