View previous topic :: View next topic |
Author |
Message |
Goverp Advocate
Joined: 07 Mar 2007 Posts: 2167
|
Posted: Tue Aug 22, 2023 1:01 pm Post subject: make error: stack frame size (2056) exceeds limit (2048) |
|
|
Making kernel-6.4.11 with the clang toolchain gives an error message:
Code: | error: stack frame size (2056) exceeds limit (2048) |
I can eliminate the message by increasing CONFIG_FRAME_WARN to say 4096. But is that the right thing to do?
<edit>I stand corrected. Changing that has no effect, I still get the same rude message, and a broken compile. It's in the depths of the amdgpu DC driver.</edit>
My guess is that this is just a warning to kernel developers to make their code smaller, prettier or more fragrant or something, rather than an actual error, but clang being clang, they're treating warnings as errors, so the make fails.
(Aside: kernel make with clang and LTO is still broken. Some fixes have come through, so IIUC clang without LTO can now handle the latest round of mitigations, but with LTO lld complains "can't move the location counter backwards" and "SRSO function pair won't alias") _________________ Greybeard |
|
Back to top |
|
|
Drako n00b
Joined: 03 Sep 2022 Posts: 9
|
Posted: Tue Aug 22, 2023 6:00 pm Post subject: |
|
|
Are you using Full LTO, ThinLTO, or neither? I was having the same "stack frame size" warning with kernel version 6.4.9 while using Full LTO. I now use ThinLTO for that reason and a couple others.
Though, in 6.4.10, the warning is gone for both Full LTO and ThinLTO. I've yet to try out 6.4.11. |
|
Back to top |
|
|
Goverp Advocate
Joined: 07 Mar 2007 Posts: 2167
|
Posted: Wed Aug 23, 2023 5:50 am Post subject: |
|
|
Thanks for the info Drako. I tried FullLTO some time ago, and the resulting kernels were much larger and slower to make, so I've only used ThinLTO for ages.
I have the same problem on 6.4.10. If you're not getting the problem it may be that you aren't using amdgpu and DC, as that's the module where the problem hits.
Thinking about it, the stack frame error message has been around for me for some time - perhaps all of 6.4, possibly even 6.3, but previously was a warning, not an error. Perhaps it's down to clang releases, as they've tightened up on a number of errors that were treated as warnings. _________________ Greybeard |
|
Back to top |
|
|
Drako n00b
Joined: 03 Sep 2022 Posts: 9
|
Posted: Wed Aug 23, 2023 3:51 pm Post subject: |
|
|
Actually, I'm using an AMD GPU and CPU; And kernel 6.4.11 compiles just fine.
Feel free to examine my kernel config: https://0x0.st/H9Zy.txt |
|
Back to top |
|
|
Goverp Advocate
Joined: 07 Mar 2007 Posts: 2167
|
Posted: Wed Aug 23, 2023 6:21 pm Post subject: |
|
|
Ah well, 6.4.12 cures it for me. _________________ Greybeard |
|
Back to top |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
|