Skip to contentSkip to conversation

Oldest still supported kernel

LTP Petr Vorel 5 participants 6 messages May 16, 2019

lore
Share & tools
lore

Petr Vorel

author
#0 lore raw

Hi,

do we declare oldest still supported kernel version?
According to distrowatch [1] still supported RHEL-6.9 is using v2.6.32. Is it the oldest version we want to support or even something older? Latest still supported stable kernel is 3.16, which is obviously too new for enterprise distros.

It'd be nice to specify latest supported kernel (and glibc) version in docs and remove old autotools checks (we have some checks for v2.6.25).

Kind regards,
Petr

[1] https://distrowatch.com/table.php?distribution=redhat

#1 lore raw

----- Original Message -----

Quoted 4 lines

Hi,

do we declare oldest still supported kernel version? According to distrowatch [1] still supported RHEL-6.9 is using v2.6.32.

6.9 is already EOL, but there still are other supported RHEL6 streams (up to 2024).

Quoted 1 line

Is it the oldest version we want to support or even something older?

I'd like minimum to be at least 3.10.0 / glibc-2.17 (RHEL7).

Older distros use LTP mostly for regression tests, so it might be acceptable for users to switch to older release tag, rather than always latest master.

There's also an option, we create a "legacy" branch for old distros, and accept only critical fixes (no new tests, rewrites, etc.). It would be unsupported, but provide place where legacy users can cooperate.

Quoted 11 lines

Latest still supported stable kernel is 3.16, which is obviously too new for enterprise distros.

It'd be nice to specify latest supported kernel (and glibc) version in docs and remove old autotools checks (we have some checks for v2.6.25).

Kind regards,
Petr

[1] https://distrowatch.com/table.php?distribution=redhat

#2 lore raw

Hi!

Quoted 10 lines
Quoted 1 line

Is it the oldest version we want to support or even something older?

I'd like minimum to be at least 3.10.0 / glibc-2.17 (RHEL7).

Older distros use LTP mostly for regression tests, so it might be acceptable for users to switch to older release tag, rather than always latest master.

There's also an option, we create a "legacy" branch for old distros, and accept only critical fixes (no new tests, rewrites, etc.). It would be unsupported, but provide place where legacy users can cooperate.

I was trying to avoid having several active branches for LTP for several reasons. Mainly to avoid people running old LTP on reasonably modern kernels because they were under an impression that older release is more stable. Hence I would like to avoid having this if possible.

Signature · 2 lines
Cyril Hrubis
[email protected]
#3 lore raw
Quoted 5 lines · Cyril Hrubis

Hi!

Quoted 3 lines
Quoted 1 line

Is it the oldest version we want to support or even something older?

I'd like minimum to be at least 3.10.0 / glibc-2.17 (RHEL7).

This minimum looks good to me.

Quoted 6 lines
Quoted 2 lines

Older distros use LTP mostly for regression tests, so it might be

acceptable

Quoted 1 line

for users to switch to older release tag, rather than always latest

master.

I think so. To switch to older release is a better option in that situation.

Quoted 10 lines
Quoted 4 lines

There's also an option, we create a "legacy" branch for old distros, and accept only critical fixes (no new tests, rewrites, etc.). It would be unsupported, but provide place where legacy users can cooperate.

I was trying to avoid having several active branches for LTP for several reasons. Mainly to avoid people running old LTP on reasonably modern kernels because they were under an impression that older release is more stable. Hence I would like to avoid having this if possible.

Yes, to maintain an old LTP branch will also cost more energy, I agree to avoid do that too.

But one more question, if a person posts a patch to fix an older issue which conflicts with the new kernel stuff, what should we do for that?

Signature · 5 lines
Regards,
Li Wang
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20190516/3c50bc3d/attachment.html>
#4 lore raw

Hi

Quoted 37 lines

On Thu, May 16, 2019 at 5:07 PM Cyril Hrubis <[email protected] <mailto:[email protected]>> wrote:

    Hi!
    > > Is it the oldest version we want to support or even something
    older?
    >
    > I'd like minimum to be at least 3.10.0 / glibc-2.17 (RHEL7).

This minimum looks good to me.

Quoted 2 lines

Older distros use LTP mostly for regression tests, so it might

be acceptable
> for users to switch to older release tag, rather than always latest master.

I think so. To switch to older release is a better option in that situation.

Quoted 3 lines

There's also an option, we create a "legacy" branch for old distros, and accept only critical fixes (no new tests, rewrites, etc.).

It would
> be unsupported, but provide place where legacy users can cooperate.

I was trying to avoid having several active branches for LTP for several
reasons. Mainly to avoid people running old LTP on reasonably modern kernels because they were under an impression that older release is more
stable. Hence I would like to avoid having this if possible.

Quoted 5 lines

Yes, to maintain an old LTP branch will also cost more energy, I agree to avoid do that too.

But one more question, if a person posts a patch to fix an older issue which conflicts with the new kernel stuff, what should we do for that?

In this situation, I think new kernel stuff has a higher priority if
older issue is not very serious.
For Oldest supported , I don't think we must specify a kernel or glibc
oldest version.

Kind Regards
Yang Xu

Quoted 4 lines
Signature · 3 lines
-- 
Regards,
Li Wang

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20190516/8fda3284/attachment.html>

#5 lore raw

Hi!

Quoted 2 lines

For Oldest supported , I don't think we must specify a kernel or glibc
oldest version.

Keeping special cases and workarounds for things that have been release more than 15 years ago has a maintenance costs as well. It makes sense to decide on oldest supported kernel/libc versions from time to time so that we can get rid of these.

Signature · 2 lines
Cyril Hrubis
[email protected]