Re: [RFC PATCH 1/2] RISC-V: Mark existing SBI as legacy SBI.

From: Atish Patra <Atish.Patra@wdc.com>
To: "hch@infradead.org" <hch@infradead.org>
Cc: "linux-riscv@lists.infradead.org"
	<linux-riscv@lists.infradead.org>,
	"alankao@andestech.com" <alankao@andestech.com>,
	"gregkh@linuxfoundation.org" <gregkh@linuxfoundation.org>,
	"gary@garyguo.net" <gary@garyguo.net>,
	"rppt@linux.ibm.com" <rppt@linux.ibm.com>,
	"paul.walmsley@sifive.com" <paul.walmsley@sifive.com>,
	"aou@eecs.berkeley.edu" <aou@eecs.berkeley.edu>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"tglx@linutronix.de" <tglx@linutronix.de>,
	"anup@brainfault.org" <anup@brainfault.org>,
	"alexios.zavras@intel.com" <alexios.zavras@intel.com>,
	"palmer@sifive.com" <palmer@sifive.com>
Subject: Re: [RFC PATCH 1/2] RISC-V: Mark existing SBI as legacy SBI.
Date: Tue, 27 Aug 2019 20:37:27 +0000	[thread overview]
Message-ID: <ac3cfe4502090354a7c49fae277adb757ad900d5.camel@wdc.com> (raw)
In-Reply-To: <20190827140304.GA21855@infradead.org>

On Tue, 2019-08-27 at 07:03 -0700, Christoph Hellwig wrote:
> > +#define SBI_EXT_LEGACY_SET_TIMER 0x0
> > +#define SBI_EXT_LEGACY_CONSOLE_PUTCHAR 0x1
> > +#define SBI_EXT_LEGACY_CONSOLE_GETCHAR 0x2
> > +#define SBI_EXT_LEGACY_CLEAR_IPI 0x3
> > +#define SBI_EXT_LEGACY_SEND_IPI 0x4
> > +#define SBI_EXT_LEGACY_REMOTE_FENCE_I 0x5
> > +#define SBI_EXT_LEGACY_REMOTE_SFENCE_VMA 0x6
> > +#define SBI_EXT_LEGACY_REMOTE_SFENCE_VMA_ASID 0x7
> > +#define SBI_EXT_LEGACY_SHUTDOWN 0x8
> 
> As Mike said legacy is a bit of a weird name.  I think this should
> be SBI01_* or so.  And pleae align the numeric values and maybe use
> an enum.
> 
Will do.

> > +
> > +#define SBI_CALL_LEGACY(which, arg0, arg1, arg2, arg3)
> > ({             \
> >  	register uintptr_t a0 asm ("a0") = (uintptr_t)(arg0);	\
> >  	register uintptr_t a1 asm ("a1") = (uintptr_t)(arg1);	\
> >  	register uintptr_t a2 asm ("a2") = (uintptr_t)(arg2);	\
> 
> Instead of the weird inline assembly with forced register allocation,
> why not move this to pure assembly?  AFAICs this is the whole
> assembly
> code we'd need:
> 
> ENTRY(sbi01_call)
>         ecall
> END(sbi01_call)
> 

That would split the implementation between C file & assembly file for
no good reason.

How about moving everything in sbi.c and just write everything inline
assembly there.

> >  /* Lazy implementations until SBI is finalized */
> > -#define SBI_CALL_0(which) SBI_CALL(which, 0, 0, 0, 0)
> > -#define SBI_CALL_1(which, arg0) SBI_CALL(which, arg0, 0, 0, 0)
> > -#define SBI_CALL_2(which, arg0, arg1) SBI_CALL(which, arg0, arg1,
> > 0, 0)
> > -#define SBI_CALL_3(which, arg0, arg1, arg2) \
> > -		SBI_CALL(which, arg0, arg1, arg2, 0)
> > -#define SBI_CALL_4(which, arg0, arg1, arg2, arg3) \
> > -		SBI_CALL(which, arg0, arg1, arg2, arg3)
> > +#define SBI_CALL_LEGACY_0(which) SBI_CALL_LEGACY(which, 0, 0, 0,
> > 0)
> > +#define SBI_CALL_LEGACY_1(which, arg0) SBI_CALL_LEGACY(which,
> > arg0, 0, 0, 0)
> > +#define SBI_CALL_LEGACY_2(which, arg0, arg1) \
> > +		SBI_CALL_LEGACY(which, arg0, arg1, 0, 0)
> > +#define SBI_CALL_LEGACY_3(which, arg0, arg1, arg2) \
> > +		SBI_CALL_LEGACY(which, arg0, arg1, arg2, 0)
> > +#define SBI_CALL_LEGACY_4(which, arg0, arg1, arg2, arg3) \
> > +		SBI_CALL_LEGACY(which, arg0, arg1, arg2, arg3)
> 
> When you touch this anyway I'd suggest you kill these rather
> pointless wrappers as well as the comment above them.

Sure. I did not want make bigger chagnes in 1st RFC patches. But
looking at the comments, I feel it was not a good decission.

I will get rid of these wrappers and move necessary ones to sbi.c

-- 
Regards,
Atish