Skip to content
This repository has been archived by the owner on Dec 3, 2019. It is now read-only.

Kernel version less than 3.4 support #131

Open
ssskip opened this issue Nov 2, 2017 · 5 comments
Open

Kernel version less than 3.4 support #131

ssskip opened this issue Nov 2, 2017 · 5 comments

Comments

@ssskip
Copy link

ssskip commented Nov 2, 2017

Hi, Thanks for developing pyflame, if it possible to add support on kernel version less than 3.4?

PTRACE_SEIZE/PTRACE_INTERRUPT (since Linux 3.4)

@ssskip ssskip changed the title Kernel version less than 3.6 support Kernel version less than 3.4 support Nov 2, 2017
@eklitzke
Copy link
Collaborator

eklitzke commented Nov 6, 2017

Hi @ssskip,

The original version of Pyflame didn't use PTRACE_SEIZE, seizing was added by commit 143144a . The last version that did not use PTRACE_SEIZE can be fetched using the git tag v1.5.2. I am going to leave this issue open because I think that Pyflame should be able to work without PTRACE_SEIZE, but in the meantime can you try v1.5.2?

@ssskip
Copy link
Author

ssskip commented Nov 7, 2017

@eklitzke thanks for your update, so if i use pt_attach version,it’s not possible to do a runtime prof on long running python process,right?

since

Unlike PTRACE_ATTACH, PTRACE_SEIZE does not stop the process.

@ssskip
Copy link
Author

ssskip commented Nov 7, 2017

@eklitzke BTW, I think append kernel version build check can be helpful for people try to compile
pyflame on unsupport version.

@eklitzke
Copy link
Collaborator

I believe this can be done as a check in configure.ac using an AC_COMPILE_IFELSE check. Kind of hard for me to test since I don't have any kernels that old. A PR would be welcome.

@xhochy
Copy link

xhochy commented Sep 3, 2018

This is also a blocker for getting newer version into conda-forge: conda-forge/staged-recipes#6601

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants