[5.10] net/sched: act_ife: convert comma to semicolon

[5.10] net/sched: act_ife: convert comma to semicolon

[Ben Hutchings]

[-- Attachment #1: Type: text/plain, Size: 1165 bytes --]

Hello stable maintainers,

There has been a build regression for the 5.10 stable branch when both
CONFIG_NET_ACT_IFE and CONFIG_PREEMPT_RT are enabled.  This was
introduced by the backport of commit ce50039be49e ("net: sched: act_ife:
initialize struct tc_ife to fix KMSAN kernel-infoleak") in 5.10.247.

After that change, tcf_ife_dump() includes the single statement:

        opt.index = ife->tcf_index,
        opt.refcnt = refcount_read(&ife->tcf_refcnt) - ref,
        opt.bindcnt = atomic_read(&ife->tcf_bindcnt) - bind,
 
        spin_lock_bh(&ife->tcf_lock);

But with CONFIG_PREEMPT_RT enabled, spin_lock_bh() is a macro whose
expansion starts with "do", so this is a syntax error.

For 5.15-rt and newer, spin_lock_bh() is a function, and 5.4 is EOL, so
only 5.10 is affected.

Please cherry-pick commit 205305c028ad ("net/sched: act_ife: convert
comma to semicolon") to fix this for 5.10.  It should be harmless to
apply to later branches as well, of course.

Ben.

-- 
Ben Hutchings
I'm always amazed by the number of people who take up solipsism because
they heard someone else explain it. - E*Borg on alt.fan.pratchett

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [5.10] net/sched: act_ife: convert comma to semicolon

[Luis Claudio R. Goncalves]

[-- Attachment #1: Type: text/plain, Size: 2007 bytes --]

On Wed, Jan 21, 2026 at 03:39:49PM +0100, Ben Hutchings wrote:
> Hello stable maintainers,
> 
> There has been a build regression for the 5.10 stable branch when both
> CONFIG_NET_ACT_IFE and CONFIG_PREEMPT_RT are enabled.  This was
> introduced by the backport of commit ce50039be49e ("net: sched: act_ife:
> initialize struct tc_ife to fix KMSAN kernel-infoleak") in 5.10.247.
> 
> After that change, tcf_ife_dump() includes the single statement:
> 
>         opt.index = ife->tcf_index,
>         opt.refcnt = refcount_read(&ife->tcf_refcnt) - ref,
>         opt.bindcnt = atomic_read(&ife->tcf_bindcnt) - bind,
>  
>         spin_lock_bh(&ife->tcf_lock);
> 
> But with CONFIG_PREEMPT_RT enabled, spin_lock_bh() is a macro whose
> expansion starts with "do", so this is a syntax error.

Ben, we had two other instances where the v5.10-rt implementation of
spin_lock_bh() as a macro required touching the code to fix build problems:

    386242acb15e rt: fix build issue in at_hdmac
    72bf92dcdcab rt: fix build issue in be2net

As there were only two instances, it made no sense at the time backporting
the RT locking implementation from v5.15-rt or newer.

I can cherry-pick the commit you mentioned to v5.10-rt and carry it until
it hits stable. I will also discuss with the maintainers for stable RT
whether it is a good idea or not to update the locking code of v5.10-rt
(EOL in Dec 2026) or not.

Thank you again for spotting this build problem!

Luis

> For 5.15-rt and newer, spin_lock_bh() is a function, and 5.4 is EOL, so
> only 5.10 is affected.
> 
> Please cherry-pick commit 205305c028ad ("net/sched: act_ife: convert
> comma to semicolon") to fix this for 5.10.  It should be harmless to
> apply to later branches as well, of course.
> 
> Ben.
> 
> -- 
> Ben Hutchings
> I'm always amazed by the number of people who take up solipsism because
> they heard someone else explain it. - E*Borg on alt.fan.pratchett


---end quoted text---

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]