linux-sparse.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
To: Palmer Dabbelt <palmer@rivosinc.com>
Cc: linux-sparse@vger.kernel.org, Kito Cheng <kito.cheng@gmail.com>,
	linux-riscv@lists.infradead.org, mkl@pengutronix.de,
	aurelien@aurel32.net,
	Linus Torvalds <torvalds@linux-foundation.org>
Subject: Re: [PATCH v1 1/6] RISC-V: Respect -Wsparse-error for -march errors
Date: Sat, 21 May 2022 23:46:18 +0200	[thread overview]
Message-ID: <20220521214618.2i5xokersg3hhzfn@mail> (raw)
In-Reply-To: <20220402050041.21302-2-palmer@rivosinc.com>

On Fri, Apr 01, 2022 at 10:00:36PM -0700, Palmer Dabbelt wrote:
> Parsing RISC-V ISA strings is extremely complicated: there are many
> extensions, versions of extensions, versions of the ISA string rules,
> and a bunch of unwritten rules to deal with all the bugs that fell out
> of that complexity.
> 
> Rather than forcing users to see an error when the ISA string parsing
> fails, just stop parsing where we get lost.  Changes tend to end up at
> the end of the ISA string, so that's probably going to work (and if
> it doesn't there's a warning to true and clue folks in).
> 
> This does have the oddity in that "-Wsparse-error -march=..." behaves
> differently than "-march... -Wsparse-error", but that's already the case
> for "--arch=... -march=..." and "-march=... --arch=...".  Both
> "-Wsparse-error" and "--arch" are sparse-specific arguments, so they're
> probably both going to be in the same place.
> 
> diff --git a/target-riscv.c b/target-riscv.c
> index 6d9113c1..f5cc6cc3 100644
> --- a/target-riscv.c
> +++ b/target-riscv.c
> @@ -60,7 +61,18 @@ static void parse_march_riscv(const char *arg)
>  			goto ext;
>  		}
>  	}
> -	die("invalid argument to '-march': '%s'\n", arg);
> +
> +unknown:
> +	/*
> +	 * This behaves like do_warn() / do_error(), but we don't have a
> +	 * position so it's just inline here.
> +	 */
> +	fflush(stdout);
> +	fprintf(stderr, "%s: invalid argument to '-march': '%s'\n",
> +		Wsparse_error == FLAG_ON ? "error" : "warning", arg);
> +	if (Wsparse_error == FLAG_ON)
> +		has_error |= ERROR_CURR_PHASE;
> +	return;

I don't like this because:
1) it's way too much intimate with the way options are parsed
   (enum flag_type should stay local to options.c).
2) -Wsparse-error is a kind of hack to ignore -Werror but keep
   a way to invoke its semantic (but I don' think anyone is using it).
3) I don't think -Wsparse-error (or GCC's -Werror) should be concerned
   with the parsing of options.

I think it would be fine, for now, to always simply report a warning,
like Linus' patch (but I would prefer to just handle the correct parsing).
If reporting an error is important, then I would be happy to jut move
this code into an helper defined in "options.c".

Best regards,
-- Luc

  reply	other threads:[~2022-05-21 21:46 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-04-02  5:00 [PATCH v1 0/6] RISC-V -march handling improvements Palmer Dabbelt
2022-04-02  5:00 ` [PATCH v1 1/6] RISC-V: Respect -Wsparse-error for -march errors Palmer Dabbelt
2022-05-21 21:46   ` Luc Van Oostenryck [this message]
2022-04-02  5:00 ` [PATCH v1 2/6] RISC-V: Match GCC's semantics for multiple -march instances Palmer Dabbelt
2022-05-21 21:52   ` Luc Van Oostenryck
2022-04-02  5:00 ` [PATCH v1 3/6] RISC-V: Remove the unimplemented ISA extensions Palmer Dabbelt
2022-05-21 22:05   ` Luc Van Oostenryck
2022-04-02  5:00 ` [PATCH v1 4/6] RISC-V: Remove "g" from the extension list Palmer Dabbelt
2022-04-02  5:00 ` [PATCH v1 5/6] RISC-V: Add the Zicsr extension Palmer Dabbelt
2022-04-02  5:00 ` [PATCH v1 6/6] RISC-V: Add the Zifencei extension Palmer Dabbelt
2022-04-04 17:52 ` [PATCH v1 0/6] RISC-V -march handling improvements Marc Kleine-Budde
2022-04-04 23:15   ` Palmer Dabbelt
2022-05-21 21:23 ` Luc Van Oostenryck
2022-07-05 23:09   ` Palmer Dabbelt

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20220521214618.2i5xokersg3hhzfn@mail \
    --to=luc.vanoostenryck@gmail.com \
    --cc=aurelien@aurel32.net \
    --cc=kito.cheng@gmail.com \
    --cc=linux-riscv@lists.infradead.org \
    --cc=linux-sparse@vger.kernel.org \
    --cc=mkl@pengutronix.de \
    --cc=palmer@rivosinc.com \
    --cc=torvalds@linux-foundation.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).