[PATCH] mtd: nand: Use ULL-suffix for big u64 constant

[PATCH] mtd: nand: Use ULL-suffix for big u64 constant

[Geert Uytterhoeven]

drivers/mtd/nand/nand_timings.c:45: warning: integer constant is too large for ‘long’ type

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
---
 drivers/mtd/nand/nand_timings.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/mtd/nand/nand_timings.c b/drivers/mtd/nand/nand_timings.c
index 8b36253420fa..e81470a8ac67 100644
--- a/drivers/mtd/nand/nand_timings.c
+++ b/drivers/mtd/nand/nand_timings.c
@@ -42,7 +42,7 @@ static const struct nand_sdr_timings onfi_sdr_timings[] = {
 		.tRHZ_max = 200000,
 		.tRLOH_min = 0,
 		.tRP_min = 50000,
-		.tRST_max = 250000000000,
+		.tRST_max = 250000000000ULL,
 		.tWB_max = 200000,
 		.tRR_min = 40000,
 		.tWC_min = 100000,
-- 
1.9.1

Re: [PATCH] mtd: nand: Use ULL-suffix for big u64 constant

[Brian Norris]

On Sat, Aug 09, 2014 at 07:07:53PM +0200, Geert Uytterhoeven wrote:
> drivers/mtd/nand/nand_timings.c:45: warning: integer constant is too large for ‘long’ type

On what compiler / static analyzer?

> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
> ---
>  drivers/mtd/nand/nand_timings.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/mtd/nand/nand_timings.c b/drivers/mtd/nand/nand_timings.c
> index 8b36253420fa..e81470a8ac67 100644
> --- a/drivers/mtd/nand/nand_timings.c
> +++ b/drivers/mtd/nand/nand_timings.c
> @@ -42,7 +42,7 @@ static const struct nand_sdr_timings onfi_sdr_timings[] = {
>  		.tRHZ_max = 200000,
>  		.tRLOH_min = 0,
>  		.tRP_min = 50000,
> -		.tRST_max = 250000000000,
> +		.tRST_max = 250000000000ULL,

See [1] and [2]. I'm pretty sure it is a bug in your tool to warn about
this. The C standard seems pretty clear that the large literal constant
is automatically promoted to a sufficiently large type, so AIUI there
should be no need for such a warning. A standards-compliant compiler
will do the right thing.

It really seems like this type of warning is only appropriate where a
smaller literal constant is immediately operated on such that it will
overflow. e.g.:

		.tRST_max = 250 * 1000 * 1000 * 1000,

But this usually results in a different type of (correct) warning, like:

  warning: integer overflow in expression [-Woverflow]

However, I could be convinced to apply this, if only to satisfy broken
tools and to avoid addressing this question over and over...

>  		.tWB_max = 200000,
>  		.tRR_min = 40000,
>  		.tWC_min = 100000,

Brian

[1] http://lists.infradead.org/pipermail/linux-mtd/2014-July/054750.html
[2] http://lists.infradead.org/pipermail/linux-arm-kernel/2014-April/252494.html

Re: [PATCH] mtd: nand: Use ULL-suffix for big u64 constant

[Geert Uytterhoeven]

Hi Brian,

On Mon, Aug 11, 2014 at 7:50 PM, Brian Norris
<computersforpeace@gmail.com> wrote:
> On Sat, Aug 09, 2014 at 07:07:53PM +0200, Geert Uytterhoeven wrote:
>> drivers/mtd/nand/nand_timings.c:45: warning: integer constant is too large for ‘long’ type
>
> On what compiler / static analyzer?

Any 32-bit version of gcc at the standard warning level, at least
until gcc 4.4.7.
With gcc 4.6 it's indeed gone.

>> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
>> ---
>>  drivers/mtd/nand/nand_timings.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/mtd/nand/nand_timings.c b/drivers/mtd/nand/nand_timings.c
>> index 8b36253420fa..e81470a8ac67 100644
>> --- a/drivers/mtd/nand/nand_timings.c
>> +++ b/drivers/mtd/nand/nand_timings.c
>> @@ -42,7 +42,7 @@ static const struct nand_sdr_timings onfi_sdr_timings[] = {
>>               .tRHZ_max = 200000,
>>               .tRLOH_min = 0,
>>               .tRP_min = 50000,
>> -             .tRST_max = 250000000000,
>> +             .tRST_max = 250000000000ULL,
>
> See [1] and [2]. I'm pretty sure it is a bug in your tool to warn about

Yes, I saw those, after I had created my patch. I decided to send it anyway ;-)

> this. The C standard seems pretty clear that the large literal constant
> is automatically promoted to a sufficiently large type, so AIUI there
> should be no need for such a warning. A standards-compliant compiler
> will do the right thing.

Actually I didn't know there existed versions of gcc that did _not_ give
this warning. This seems to be a recent thing on gcc for 32-bit targets.
For 64-bit targets, apparently it only gives a warning if the constant
is that large that it can no longer fit in a 64-bit signed integer.

> However, I could be convinced to apply this, if only to satisfy broken
> tools and to avoid addressing this question over and over...

So please apply. Too many people are still using 32-bit targets and
gcc < 4.6.

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: [PATCH] mtd: nand: Use ULL-suffix for big u64 constant

[Brian Norris]

On Mon, Aug 11, 2014 at 08:51:51PM +0200, Geert Uytterhoeven wrote:
> On Mon, Aug 11, 2014 at 7:50 PM, Brian Norris
> <computersforpeace@gmail.com> wrote:
> > On Sat, Aug 09, 2014 at 07:07:53PM +0200, Geert Uytterhoeven wrote:
> >> drivers/mtd/nand/nand_timings.c:45: warning: integer constant is too large for ‘long’ type
> >
> > On what compiler / static analyzer?
> 
> Any 32-bit version of gcc at the standard warning level, at least
> until gcc 4.4.7.
> With gcc 4.6 it's indeed gone.
> 
> >> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
> >> ---
> >>  drivers/mtd/nand/nand_timings.c | 2 +-
> >>  1 file changed, 1 insertion(+), 1 deletion(-)
> >>
> >> diff --git a/drivers/mtd/nand/nand_timings.c b/drivers/mtd/nand/nand_timings.c
> >> index 8b36253420fa..e81470a8ac67 100644
> >> --- a/drivers/mtd/nand/nand_timings.c
> >> +++ b/drivers/mtd/nand/nand_timings.c
> >> @@ -42,7 +42,7 @@ static const struct nand_sdr_timings onfi_sdr_timings[] = {
> >>               .tRHZ_max = 200000,
> >>               .tRLOH_min = 0,
> >>               .tRP_min = 50000,
> >> -             .tRST_max = 250000000000,
> >> +             .tRST_max = 250000000000ULL,
> >
> > See [1] and [2]. I'm pretty sure it is a bug in your tool to warn about
> 
> Yes, I saw those, after I had created my patch. I decided to send it anyway ;-)
> 
> > this. The C standard seems pretty clear that the large literal constant
> > is automatically promoted to a sufficiently large type, so AIUI there
> > should be no need for such a warning. A standards-compliant compiler
> > will do the right thing.
> 
> Actually I didn't know there existed versions of gcc that did _not_ give
> this warning. This seems to be a recent thing on gcc for 32-bit targets.
> For 64-bit targets, apparently it only gives a warning if the constant
> is that large that it can no longer fit in a 64-bit signed integer.

I'm curious: do your broken compilers also generate broken code/data?
i.e., does gcc 4.4.7 give 250000000000 its proper ULL value still?

> > However, I could be convinced to apply this, if only to satisfy broken
> > tools and to avoid addressing this question over and over...
> 
> So please apply. Too many people are still using 32-bit targets and
> gcc < 4.6.

OK, you've convinced me ;) Applied to l2-mtd.git/next, with an editorial
comment.

Brian

Re: [PATCH] mtd: nand: Use ULL-suffix for big u64 constant

[Geert Uytterhoeven]

Hi Brian,

On Mon, Aug 11, 2014 at 9:06 PM, Brian Norris
<computersforpeace@gmail.com> wrote:
> I'm curious: do your broken compilers also generate broken code/data?
> i.e., does gcc 4.4.7 give 250000000000 its proper ULL value still?

Yes it does (checked the assembler output).

>> > However, I could be convinced to apply this, if only to satisfy broken
>> > tools and to avoid addressing this question over and over...
>>
>> So please apply. Too many people are still using 32-bit targets and
>> gcc < 4.6.
>
> OK, you've convinced me ;) Applied to l2-mtd.git/next, with an editorial
> comment.

Thanks!

BTW, I had some faint memories of integer constants never being 64-bit
without the "LL" suffix, but that's no longer true in c99 (I did check that a
few days ago).

The warning does go away when adding -std=c99 with gcc-4.4.
Interestingly, it doesn't appear with -std=c89 with gcc 4.6.2 or 4.8.2.

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds