Re: [PATCH v2 06/18] m68k: Replace setup_irq() by request_irq()

[Finn Thain <fthain@telegraphics.com.au>]

On Thu, 27 Feb 2020, Greg Ungerer wrote:

> On 27/2/20 8:31 am, Finn Thain wrote:
> 
> >>>
> >>> BTW, one of the benefits of "%s: request_irq failed" is that a 
> >>> compilation unit with multiple request_irq calls permits the 
> >>> compiler to coalesce all duplicated format strings. Whereas, that's 
> >>> not possible with "foo: request_irq failed" and "bar: request_irq 
> >>> failed".
> >>
> >> Given the wide variety of message text used with failed request_irq() 
> >> calls it would be shear luck that this matched anything else. A quick 
> >> grep shows that "%s: request_irq() failed\n" has no other exact 
> >> matches in the current kernel source.
> > 
> > You are overlooking the patches in this series that produce multiple 
> > identical format strings.
> 
> No I didn't :-)  None of these will end up compiled in at the same time. 
> The various ColdFire SoC parts have a single timer hardware module - and 
> only the required one will be compiled in, not all of them.
> 

I was referring to e.g. [PATCH v2 08/18] MIPS: Replace setup_irq() by 
request_irq(), in which you can find this:

@@ -116,8 +110,16 @@ static void __init ar7_irq_init(int base)
                                                 handle_level_irq);
        }
 
-       setup_irq(2, &ar7_cascade_action);
-       setup_irq(ar7_irq_base, &ar7_cascade_action);
+       if (request_irq(2, no_action, IRQF_NO_THREAD, "AR7 cascade interrupt",
+                       NULL)) {
+               pr_err("%s: request_irq() failed\n",
+                      "AR7 cascade interrupt");
+       }
+       if (request_irq(ar7_irq_base, no_action, IRQF_NO_THREAD,
+                       "AR7 cascade interrupt", NULL)) {
+               pr_err("%s: request_irq() failed\n",
+                      "AR7 cascade interrupt");
+       }
        set_c0_status(IE_IRQ0);
 }
 
BTW, I think that deduplication of string constants can happen during LTO, 
so the benefit of consistency need not be confined to a compilation unit. 
I don't think this is relevant to kernel builds.