Re: [PATCH v2 05/12] x86: rework arch_local_irq_restore() to not use popf

[Andy Lutomirski <luto@xxxxxxxxxxxxxx>]

> On Nov 22, 2020, at 9:22 PM, Jürgen Groß <jgross@xxxxxxxx> wrote:
> 
> On 22.11.20 22:44, Andy Lutomirski wrote:
>>> On Sat, Nov 21, 2020 at 10:55 PM Jürgen Groß <jgross@xxxxxxxx> wrote:
>>> 
>>> On 20.11.20 12:59, Peter Zijlstra wrote:
>>>> On Fri, Nov 20, 2020 at 12:46:23PM +0100, Juergen Gross wrote:
>>>>> +static __always_inline void arch_local_irq_restore(unsigned long flags)
>>>>> +{
>>>>> +    if (!arch_irqs_disabled_flags(flags))
>>>>> +            arch_local_irq_enable();
>>>>> +}
>>>> 
>>>> If someone were to write horrible code like:
>>>> 
>>>>       local_irq_disable();
>>>>       local_irq_save(flags);
>>>>       local_irq_enable();
>>>>       local_irq_restore(flags);
>>>> 
>>>> we'd be up some creek without a paddle... now I don't _think_ we have
>>>> genius code like that, but I'd feel saver if we can haz an assertion in
>>>> there somewhere...
>>>> 
>>>> Maybe something like:
>>>> 
>>>> #ifdef CONFIG_DEBUG_ENTRY // for lack of something saner
>>>>       WARN_ON_ONCE((arch_local_save_flags() ^ flags) & X86_EFLAGS_IF);
>>>> #endif
>>>> 
>>>> At the end?
>>> 
>>> I'd like to, but using WARN_ON_ONCE() in include/asm/irqflags.h sounds
>>> like a perfect receipt for include dependency hell.
>>> 
>>> We could use a plain asm("ud2") instead.
>> How about out-of-lining it:
>> #ifdef CONFIG_DEBUG_ENTRY
>> extern void warn_bogus_irqrestore();
>> #endif
>> static __always_inline void arch_local_irq_restore(unsigned long flags)
>> {
>>        if (!arch_irqs_disabled_flags(flags)) {
>>                arch_local_irq_enable();
>>        } else {
>> #ifdef CONFIG_DEBUG_ENTRY
>>                if (unlikely(arch_local_irq_save() & X86_EFLAGS_IF))
>>                     warn_bogus_irqrestore();
>> #endif
>> }
> 
> This couldn't be a WARN_ON_ONCE() then (or it would be a catch all).

If you put the WARN_ON_ONCE in the out-of-line helper, it should work 
reasonably well.

> Another approach might be to open-code the WARN_ON_ONCE(), like:
> 
> #ifdef CONFIG_DEBUG_ENTRY
> extern void warn_bogus_irqrestore(bool *once);
> #endif
> 
> static __always_inline void arch_local_irq_restore(unsigned long flags)
> {
>    if (!arch_irqs_disabled_flags(flags))
>        arch_local_irq_enable();
> #ifdef CONFIG_DEBUG_ENTRY
>    {
>        static bool once;
> 
>        if (unlikely(arch_local_irq_save() & X86_EFLAGS_IF))
>            warn_bogus_irqrestore(&once);
>    }
> #endif
> }
> 

I don’t know precisely what a static variable in an __always_inline function 
will do, but I imagine it will be, at best, erratic, especially when modules 
are involved.

> 
> Juergen
> <OpenPGP_0xB0DE9DD628BF132F.asc>