From: John Garry <john.garry@huawei.com>
To: Johannes Berg <johannes@sipsolutions.net>,
Joe Perches <joe@perches.com>,
Andrew Morton <akpm@linux-foundation.org>,
Andy Shevchenko <andy.shevchenko@gmail.com>,
Kalle Valo <kvalo@codeaurora.org>, <jakub.kicinski@netronome.com>,
<yamada.masahiro@socionext.com>, "Arnd Bergmann" <arnd@arndb.de>,
<viro@zeniv.linux.org.uk>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Cc: <linux-wireless@vger.kernel.org>, <nbd@nbd.name>
Subject: Re: Question on FIELD_PREP() for static array
Date: Thu, 11 Oct 2018 15:24:48 +0100 [thread overview]
Message-ID: <830ff9fb-07b1-86b6-4f57-af549a37a8c0@huawei.com> (raw)
In-Reply-To: <1539195207.3687.178.camel@sipsolutions.net>
On 10/10/2018 19:13, Johannes Berg wrote:
> On Wed, 2018-10-10 at 10:33 -0700, Joe Perches wrote:
>>
>>> Specifically it doesn't like the __BF_FIELD_CHECK() in FIELD_PREP().
>>>
>>> Any ideas on compiler trickery we could do with the FIELD_PREP()
>>> definition to avoid this issue (i.e. enforce the check but only use the
>>> constant value)?
>>
thanks guys
>> Perhaps __bf_shf should not use __builtin_ffsll.
>
> __bf_shf() is a constant expression, and is fine in this context.
>
> The problem is the use of the compound statement here:
>
> static int x[2] = {
> ({ (void)(0); 1; }),
> 0,
> }
>
> similarly fails to compile.
>
> I've recently run into a similar situation, namely in
> include/net/netlink.h, and the applicable way to solve it here would be
> something like this:
>
> diff --git a/include/linux/bitfield.h b/include/linux/bitfield.h
> index 3f1ef4450a7c..0680d641923f 100644
> --- a/include/linux/bitfield.h
> +++ b/include/linux/bitfield.h
> @@ -49,19 +49,16 @@
>
> #define __bf_shf(x) (__builtin_ffsll(x) - 1)
>
> +#define BUILD_BUG_ON_RET_ZERO(cond) (sizeof(char[1 - 2*!!(cond)]) - 1)
> +#define BUILD_BUG_ON_NOT_POW2_RET_ZERO(n) BUILD_BUG_ON_RET_ZERO(((n) & ((n) - 1)) != 0)
> +
> #define __BF_FIELD_CHECK(_mask, _reg, _val, _pfx) \
> - ({ \
> - BUILD_BUG_ON_MSG(!__builtin_constant_p(_mask), \
> - _pfx "mask is not constant"); \
> - BUILD_BUG_ON_MSG((_mask) == 0, _pfx "mask is zero"); \
> - BUILD_BUG_ON_MSG(__builtin_constant_p(_val) ? \
> - ~((_mask) >> __bf_shf(_mask)) & (_val) : 0, \
> - _pfx "value too large for the field"); \
> - BUILD_BUG_ON_MSG((_mask) > (typeof(_reg))~0ull, \
> - _pfx "type of reg too small for mask"); \
> - __BUILD_BUG_ON_NOT_POWER_OF_2((_mask) + \
> - (1ULL << __bf_shf(_mask))); \
> - })
> + BUILD_BUG_ON_RET_ZERO(!__builtin_constant_p(_mask)) + \
> + BUILD_BUG_ON_RET_ZERO((_mask) == 0) + \
> + BUILD_BUG_ON_RET_ZERO(__builtin_constant_p(_val) ? \
> + ~((_mask) >> __bf_shf(_mask)) & (_val) : 0) + \
> + BUILD_BUG_ON_RET_ZERO((_mask) > (typeof(_reg))~0ull) + \
> + BUILD_BUG_ON_NOT_POW2_RET_ZERO((_mask) + (1ULL << __bf_shf(_mask)))
>
> /**
> * FIELD_FIT() - check if value fits in the field
> @@ -85,10 +82,8 @@
> * be combined with other fields of the bitfield using logical OR.
> */
> #define FIELD_PREP(_mask, _val) \
> - ({ \
> - __BF_FIELD_CHECK(_mask, 0ULL, _val, "FIELD_PREP: "); \
> - ((typeof(_mask))(_val) << __bf_shf(_mask)) & (_mask); \
> - })
> + (__BF_FIELD_CHECK(_mask, 0ULL, _val, "FIELD_PREP: ") + \
> + (((typeof(_mask))(_val) << __bf_shf(_mask)) & (_mask)))
>
> /**
> * FIELD_GET() - extract a bitfield element
>
>
> Note that this is an incomplete patch - everything but FIELD_PREP will
> not compile with this.
>
> Also, BUILD_BUG_ON_RET_ZERO and BUILD_BUG_ON_NOT_POW2_RET_ZERO should
> probably have better names, or perhaps do the positive way that I did in
> __NLA_ENSURE, e.g. CONST_ASSERT()/CONST_ASSERT_IS_POWER_OF_2()? I guess
> they should go to build_bug.h as well...
Seems reasonable. However I did try this and was getting compiler
warnings about VLA, from a non-constant being fed into
BUILD_BUG_ON_RET_ZERO(), related to sizeof char[]:
drivers/iio/adc/meson_saradc.c:375:2: warning: ISO C90 forbids variable
length array [-Wvla]
regval = FIELD_PREP(MESON_SAR_ADC_CHAN_LIST_ENTRY_MASK(0),
Surely __NLA_ENSURE is getting a similar issue as it uses a similar
principle, no? I see that this is in -next now, but could not this macro
or derivatives being referenced.
>
Much appreciated,
John
> johannes
>
> .
>
next prev parent reply other threads:[~2018-10-11 14:25 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-10-10 17:16 John Garry
2018-10-10 17:33 ` Joe Perches
2018-10-10 18:13 ` Johannes Berg
2018-10-11 14:24 ` John Garry [this message]
2018-10-11 15:23 ` Johannes Berg
2018-10-11 16:16 ` John Garry
2018-10-11 17:26 ` John Garry
2018-10-11 19:13 ` Johannes Berg
2018-10-11 19:22 ` Johannes Berg
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=830ff9fb-07b1-86b6-4f57-af549a37a8c0@huawei.com \
--to=john.garry@huawei.com \
--cc=akpm@linux-foundation.org \
--cc=andy.shevchenko@gmail.com \
--cc=arnd@arndb.de \
--cc=jakub.kicinski@netronome.com \
--cc=joe@perches.com \
--cc=johannes@sipsolutions.net \
--cc=kvalo@codeaurora.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-wireless@vger.kernel.org \
--cc=nbd@nbd.name \
--cc=viro@zeniv.linux.org.uk \
--cc=yamada.masahiro@socionext.com \
/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
all inboxes | Powered by JetHome®