linux-kernel - Re: [PATCH] ata: ahci: fix enum constants for gcc-13

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

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Thu, 1 Dec 2022 12:48:40 -0800
From:   Randy Dunlap <rdunlap@...radead.org>
To:     Arnd Bergmann <arnd@...nel.org>,
        Damien Le Moal <damien.lemoal@...nsource.wdc.com>
Cc:     linux-kernel@...r.kernel.org, Arnd Bergmann <arnd@...db.de>,
        Luis Machado <luis.machado@....com>, linux-ide@...r.kernel.org,
        stable@...r.kernel.org
Subject: Re: [PATCH] ata: ahci: fix enum constants for gcc-13



On 12/1/22 12:43, Arnd Bergmann wrote:
> From: Arnd Bergmann <arnd@...db.de>
> 
> gcc-13 slightly changes the type of constant expressions that are deifined

                                                                    defined

> in an enum, which triggers a compile time sanity check in libata:
> 
> linux/drivers/ata/libahci.c: In function 'ahci_led_store':
> linux/include/linux/compiler_types.h:357:45: error: call to '__compiletime_assert_302' declared with attribute error: BUILD_BUG_ON failed: sizeof(_s) > sizeof(long)
> 357 | _compiletime_assert(condition, msg, __compiletime_assert_, __COUNTER__)
> 
> The new behavior is that sizeof() returns the same value for the
> constant as it does for the enum type, which is generally more sensible
> and consistent.
> 
> The problem in libata is that it contains a single enum definition for
> lots of unrelated constants, some of which are large positive (unsigned)
> integers like 0xffffffff, while others like (1<<31) are interpreted as
> negative integers, and this forces the enum type to become 64 bit wide
> even though most constants would still fit into a signed 32-bit 'int'.
> 
> Fix this by changing the entire enum definition to use BIT(x) in place
> of (1<<x), which results in all values being seen as 'unsigned' and
> fitting into an unsigned 32-bit type.
> 
> Link: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107917
> Link: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107405
> Reported-by: Luis Machado <luis.machado@....com>
> Cc: linux-ide@...r.kernel.org
> Cc: Damien Le Moal <damien.lemoal@...nsource.wdc.com>
> Cc: stable@...r.kernel.org
> Signed-off-by: Arnd Bergmann <arnd@...db.de>
> ---
> Luis, I don't have gcc-13 installed on the machine I used for
> creating this patch, can you give this a spin and see if it
> addresses the build failure?
> ---
>  drivers/ata/ahci.h | 234 ++++++++++++++++++++++-----------------------
>  1 file changed, 117 insertions(+), 117 deletions(-)

What #include <linux/bits.h> ?
or is it just done indirectly?

thanks.
-- 
~Randy