Re: [GIT pull] x86/urgent for 5.3-rc5

From: Linus Torvalds <torvalds@linux-foundation.org>
To: Borislav Petkov <bp@suse.de>
Cc: Pu Wen <puwen@hygon.cn>, Thomas Gleixner <tglx@linutronix.de>,
	Tom Lendacky <thomas.lendacky@amd.com>,
	Linux List Kernel Mailing <linux-kernel@vger.kernel.org>,
	"the arch/x86 maintainers" <x86@kernel.org>
Subject: Re: [GIT pull] x86/urgent for 5.3-rc5
Date: Tue, 27 Aug 2019 09:55:08 -0700	[thread overview]
Message-ID: <CAHk-=wj_E58JskechbJyWwpzu5rwKFHEABr4dCZjS+JBvv67Uw@mail.gmail.com> (raw)
In-Reply-To: <20190826125342.GC28610@zn.tnic>

On Mon, Aug 26, 2019 at 5:53 AM Borislav Petkov <bp@suse.de> wrote:
>
> +       if (!changed) {
> +               pr_emerg(
> +"RDRAND gives funky smelling output, might consider not using it by booting with \"nordrand\"");
> +               WARN_ON_ONCE(1);
> +       }

Side note: I'd suggest

        if (WARN_ON_ONCE(!changed))
                pr_emerg("RDRAND gives funky smelling output, might
consider not using it by booting with \"nordrand\"");

instead.

Note that WARN_ON_ONCE() will only _warn_ once, but it always returns
the conditional.

So it's equivalent to your version ("always print, warn once") except for:

 (a) smaller
 (b) different order of warning and pr_emerg()
 (c) the WARN_ON_ONCE() also contains the proper "unlikely()", so it
generates better code

and I just try to encourage people to use that "if (WARN_ON())" format
for these kinds of "shouldn't happen, so warn and then do something".

Although usually the "do something" is just "return and refuse to do
anything further" rather than the additional printk.

               Linus