linux-kernel - Re: [PATCH v2] riscv: lib: Check if output in asm goto supported

Open Source and information security mailing list archives

Message-ID: <ZaqT2E0rH3UbdnV0@ghost>
Date: Fri, 19 Jan 2024 07:23:36 -0800
From: Charlie Jenkins <charlie@...osinc.com>
To: Andrew Jones <ajones@...tanamicro.com>
Cc: Paul Walmsley <paul.walmsley@...ive.com>,
	Palmer Dabbelt <palmer@...belt.com>,
	Albert Ou <aou@...s.berkeley.edu>, linux-riscv@...ts.infradead.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] riscv: lib: Check if output in asm goto supported

On Fri, Jan 19, 2024 at 10:41:14AM +0100, Andrew Jones wrote:
> On Thu, Jan 18, 2024 at 02:36:45PM -0800, Charlie Jenkins wrote:
> > The output field of an asm goto statement is not supported by all
> > compilers. If it is not supported, fallback to the non-optimized code.
> > 
> > Signed-off-by: Charlie Jenkins <charlie@...osinc.com>
> > Fixes: a04c192eabfb ("riscv: Add checksum library")
> > ---
> > The OutputOperands field for asm goto statements is only supported
> > starting from GCC 11. Split the asm goto to remove the use of this
> > feature.
> > ---
> > Changes in v2:
> > - Use CC_HAS_ASM_GOTO_TIED_OUTPUT
> > - Link to v1: https://lore.kernel.org/r/20240118-csum_remove_output_operands_asm_goto-v1-1-47c672bb9d4b@rivosinc.com
> > ---
> >  arch/riscv/lib/csum.c | 2 ++
> >  1 file changed, 2 insertions(+)
> > 
> > diff --git a/arch/riscv/lib/csum.c b/arch/riscv/lib/csum.c
> > index 06ce8e7250d9..af3df5274ccb 100644
> > --- a/arch/riscv/lib/csum.c
> > +++ b/arch/riscv/lib/csum.c
> > @@ -156,6 +156,7 @@ do_csum_with_alignment(const unsigned char *buff, int len)
> >  	end = (const unsigned long *)(buff + len);
> >  	csum = do_csum_common(ptr, end, data);
> >  
> > +#ifdef CC_HAS_ASM_GOTO_TIED_OUTPUT
> 
> Can't we just add another IS_ENABLED() to the if rather than this #ifdef?

Unfortunately no. GCC throws syntax before it determines if a branch
will never be taken, so even though the code is not emitted it will
still fail with IS_ENABLED.

> 
> >  	/*
> >  	 * Zbb support saves 6 instructions, so not worth checking without
> >  	 * alternatives if supported
> > @@ -214,6 +215,7 @@ do_csum_with_alignment(const unsigned char *buff, int len)
> >  		return csum >> 16;
> >  	}
> >  no_zbb:
> > +#endif /* CC_HAS_ASM_GOTO_TIED_OUTPUT */
> >  #ifndef CONFIG_32BIT
> >  	csum += ror64(csum, 32);
> >  	csum >>= 32;
> 
> BTW, I wonder how/if the check for CC_HAS_ASM_GOTO_TIED_OUTPUT in
> init/Kconfig is working as expected. I see $CC, as opposed to $(CC),
> being used there. I believe $CC is just the expansion of $C with a
> 'C' appended.

Huh that is strange. It does work but I am not sure how.

- Charlie

> 
> Thanks,
> drew
> 
> > 
> > ---
> > base-commit: 080c4324fa5e81ff3780206a138223abfb57a68e
> > change-id: 20240118-csum_remove_output_operands_asm_goto-49922c141ce7
> > -- 
> > - Charlie
> > 
> > 
> > _______________________________________________
> > linux-riscv mailing list
> > linux-riscv@...ts.infradead.org
> > http://lists.infradead.org/mailman/listinfo/linux-riscv

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