linux-kernel - Re: [PATCH, RAW] IRQ: Maintain irq number globally rather than passing to IRQ handlers

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Fri, 06 Oct 2006 12:38:59 -0400
From:	Jeff Garzik <jeff@...zik.org>
To:	Linus Torvalds <torvalds@...l.org>
CC:	David Howells <dhowells@...hat.com>, linux-kernel@...r.kernel.org,
	linux-arch@...r.kernel.org, Ingo Molnar <mingo@...e.hu>,
	Alan Cox <alan@...rguk.ukuu.org.uk>,
	Andrew Morton <akpm@...l.org>,
	Thomas Gleixner <tglx@...utronix.de>,
	Dmitry Torokhov <dtor@...l.ru>, Greg KH <greg@...ah.com>,
	David Brownell <david-b@...bell.net>,
	Alan Stern <stern@...land.harvard.edu>
Subject: Re: [PATCH, RAW] IRQ: Maintain irq number globally rather than passing
 to IRQ handlers

Linus Torvalds wrote:
> 
> On Fri, 6 Oct 2006, Jeff Garzik wrote:
>> Here is the raw, un-split-up first pass of the irq argument removal patch
>> (500K):	http://gtf.org/garzik/misc/patch.irq-remove
> 
> So I'm not at all as sure about this as about the "regs" stuff.
> 
> The "regs" value has always been controversial. It's pretty much always 
> existed (due to the keyboard hander and the magic debugging keysequences), 
> and anybody who looks at 0.01 will quickly realize that the keyboard 
> driver was one of the very first drivers (I think it's even written in 
> assembly at that point: originally _all_ of what was to become Linux was 
> pure asm, the whole "oh, cool, I could write this part in C" came later). 
> But it's been pretty much a special case since day #1, purely for that 
> "press a key to see where the h*ck we hung" case.

Chuckle :)

> In contrast, the irq argument itself is really no different from the 
> cookie we pass in on registration - it's just passing it back to the 
> driver that requested the thing. So unlike "regs", there's not really 
> anything strange about it, and there's nothing really "wrong" with having 
> it there.

It doesn't have the colorful history of pt_regs, but the 'irq' argument 
is dead weight.  I'd say the wrongness stems from its utter uselessness.

Out of ~1100 irq handlers, the irq parameter is used in ~50.  The vast 
majority of those 50 uses are debug printks, or abused as a "did I call 
myself?" internal driver flag.  The number of "real" uses is under 15, 
and those are all ancient ISA or platform drivers that pre-date my ~10 
year history with Linux.

So, I don't see any convincing argument to keep it.  And if we are going 
to kill it, given the pt_regs churn, this is probably the best 
opportunity we'll have in years.

Another weak-but-still-present argument in favor of killing it is that 
this change would IMO future-proof irq handlers, against more exotic irq 
handling methods that may come down the pipe.

	Jeff


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/