sirlucjan/kernel-patches

Build breakage with 0005-tty-Allow-setting-the-number-of-available-virtual-TT.patch

Closed this issue · 3 comments

ms178 commented

Hello Piotr, while trying out other patches of your repo (with Vanilla sources this time), I came across an issue with the mentioned patch within the fixes-miscellaneous folder.

0005-tty-Allow-setting-the-number-of-available-virtual-TT.patch causes a compilation failure when using the make binrpm-pkg or make rpm-pkg script with the following error:

HDRINST usr/include/linux/version.h error: include/uapi/linux/vt.h: leak CONFIG_NR_TTY_DEVICES to user-space make[6]: *** [scripts/Makefile.headersinst:63: usr/include/linux/vt.h] Error 1

This was also seen in other downstream projects before which I found when researching the error message: Frogging-Family/linux-tkg#140 (comment)

Removing that patch fixes the problem.

P.S: You asked me yesterday if I could test the speculative-lru patchset, I can now confirm that the speculative-lru patches apply and build fine but the order is important, I needed to apply the lru-patches before the speculative-lru patches. The other way around doesn't apply correctly.

Yes, I know about this error. On Arch Linux everything works and that's why I keep this patch at my place. Yes, I am well aware that LRU should be applied first and then Speculative. - and it is intentional. Does UKSM no longer cause booting problems?

ms178 commented

As reported yesterday in the other issue, the new version of UKSM booted fine albeit the performance regressed a bit. Feel free to close this one, if these are known errors.

The other way was more troublesome. It is completely impossible to make it so that the order is arbitrary. Yes, the bug with the 0005-tty-Allow-setting-the-number-of-available-virtual-TT.patch is known - truthfully it only works with Arch Linux. I'm not sure about Gentoo or Slackaware - I haven't had any reports here :D