On 2018-04-07 19:06, Philippe Gerum wrote: > On 04/07/2018 07:04 PM, Jan Kiszka wrote: >> On 2018-04-07 18:58, Philippe Gerum wrote: >>> On 04/06/2018 07:21 PM, Jan Kiszka wrote: >>>> The following changes since commit >>>> 3d21f886ffae3ffdef54a19ec53393482626a84c: >>>> >>>> ipipe-core-4.9.90-x86-6 (2018-03-26 17:36:13 +0200) >>>> >>>> are available in the git repository at: >>>> >>>> git://git.xenomai.org/ipipe-jki >>>> >>>> for you to fetch changes up to 3fac4646875dbbab6969cf080a02467b578c55be: >>>> >>>> Merge tag 'v4.9.92' into for-upstream/4.9 (2018-04-06 18:45:31 +0200) >>>> >>>> ---------------------------------------------------------------- >>>> (excluding stable merge) >>>> Jan Kiszka (4): >>>> arm/ipipe: Avoid open-coded __ipipe_call_mayday >>>> ipipe: Fix root domain guard in preempt_schedule_notrace >>>> x86/ipipe: Deny JUMP_LABEL over I-pipe >>>> x86/ipipe: Disable access_ok context under I-pipe >>>> >>>> arch/arm/kernel/ipipe.c | 3 +-- >>>> arch/x86/Kconfig | 2 +- >>>> arch/x86/include/asm/uaccess.h | 2 +- >>>> kernel/sched/core.c | 2 +- >>>> 4 files changed, 4 insertions(+), 5 deletions(-) >>>> >>>> diff --git a/arch/arm/kernel/ipipe.c b/arch/arm/kernel/ipipe.c >>>> index b95c8128c5de..e7642bd28871 100644 >>>> --- a/arch/arm/kernel/ipipe.c >>>> +++ b/arch/arm/kernel/ipipe.c >>>> @@ -354,8 +354,7 @@ void __ipipe_exit_irq(struct pt_regs *regs) >>>> * so prefer test then maybe clear over >>>> * test_and_clear. >>>> */ >>>> - ipipe_clear_thread_flag(TIP_MAYDAY); >>>> - __ipipe_notify_trap(IPIPE_TRAP_MAYDAY, regs); >>>> + __ipipe_call_mayday(regs); >>>> } >>>> } >>> >>> Nack. If the purpose is to prevent open coding, then you should provide >>> a light __ipipe_call_mayday() which does not fiddle with the IRQ state >>> uselessly. __ipipe_irq_exit does run with hard IRQs off by design. >> >> You already merged this into 4.4. This is just the same thing with the >> same reasoning. >> > > Yes, but my comment made clear that it was useless. Maybe the second > submission could account for this. >
I doubt this reasoning: __ipipe_call_mayday is not hot-path. So I see no value in optimizing the - granted - unneeded hard_local_irq_save away by providing a variant of __ipipe_call_mayday that does not do that. Jan -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 181 bytes Desc: OpenPGP digital signature URL: <http://xenomai.org/pipermail/xenomai/attachments/20180407/cf4856f4/attachment.sig> _______________________________________________ Xenomai mailing list Xenomai@xenomai.org https://xenomai.org/mailman/listinfo/xenomai
