linux-kernel - Re: [PATCH] lpfc: use %zd format string for size

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] [day] [month] [year] [list]

Date:   Fri, 28 Oct 2016 16:16:48 -0700
From:   Vineet Gupta <Vineet.Gupta1@...opsys.com>
To:     Arnd Bergmann <arnd@...db.de>
CC:     arcml <linux-snps-arc@...ts.infradead.org>,
        lkml <linux-kernel@...r.kernel.org>,
        Claudiu Zissulescu <Claudiu.Zissulescu@...opsys.com>
Subject: Re: [PATCH] lpfc: use %zd format string for size_t

On 10/28/2016 03:03 PM, Arnd Bergmann wrote:
> On Friday, October 28, 2016 2:58:33 PM CEST Vineet Gupta wrote:
>> On 10/28/2016 02:52 PM, Vineet Gupta wrote:
>>> On 10/28/2016 02:44 PM, Vineet Gupta wrote:
>>>> This is configuration specific, and something caused your compiler to
>>>>> be built assuming that size_t is unsigned long, while the kernel
>>>>> headers are assuming it should be unsigned int.
>>>
>>> So yes this seems to be target specific gcc thing
>>>
>>> for ARC 4.8
>>>
>>> #define PTRDIFF_TYPE "int"
>>>
>>> ARM
>>>
>>> #ifndef PTRDIFF_TYPE
>>> #define PTRDIFF_TYPE (TARGET_AAPCS_BASED ? "int" : "long int")
>>> #endif
>>>
>>> ARC gcc 6.2
>>>
>>> #undef PTRDIFF_TYPE
>>> #define PTRDIFF_TYPE "long int"
>>
>> Actually we need to adjust SIZE_TYPE (unsigned int) and PTRDIFF_TYPE (int) in the
>> gcc 6.x to fix this issue. And that is exactly what ARC gcc 4.8 have.
> 
> What compiler versions are most commonly used these days?

gcc 4.8 is used in production, but internally we are now moving towards 6.0 (to be
officially released soon)

> You should probably stay with the version that most people have
> and then update either the compiler or the kernel, whichever
> diverges from it.

In this case, the issue is simple - gcc 6.x doesn't behave the same as 4.8 so it
needs fixing.

> I see in the gcc git log that the version that had "int" got removed
> at some point, and the version that had "unsigned int" was added
> later.

The upstream version (per initial port) always had

#define SIZE_TYPE "long unsigned int"

which we fixed out-of-tree for 4.8 and this needs to be fixed now for gcc 6.x in
upstream too.

-Vineet