linux-kernel - Re: [PATCH v1] x86/io: Define readq()/writeq() to use 64-bit type

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:   Tue, 23 Jan 2018 00:32:14 +0000
From:   "Mehta, Sohil" <sohil.mehta@...el.com>
To:     "tglx@...utronix.de" <tglx@...utronix.de>,
        "hpa@...or.com" <hpa@...or.com>,
        "andriy.shevchenko@...ux.intel.com" 
        <andriy.shevchenko@...ux.intel.com>,
        "mingo@...hat.com" <mingo@...hat.com>,
        "x86@...nel.org" <x86@...nel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
CC:     "mitake@....info.waseda.ac.jp" <mitake@....info.waseda.ac.jp>
Subject: Re: [PATCH v1] x86/io: Define readq()/writeq() to use 64-bit type

On Fri, 2018-01-19 at 16:33 +0200, Andy Shevchenko wrote:
> Since non atomic readq() and writeq() were added some of the drivers
> would like to use it in a manner of:
> 
>  #include <io-64-nonatomic-lo-hi.h>
> ...
>  pr_debug("Debug value of some register: %016llx\n", readq(addr));
> 
> However, lo_hi_readq() always returns __u64 data, while readq()
> on x86_64 defines it as unsigned long. and thus compiler warns
> about type mismatch, although they are both 64-bit on x86_64.
> 
> Convert readq() and writeq() on x86 to operate on deterministic
> 64-bit type. The most of architectures in the kernel already are
> using
> either unsigned long long, or u64 type for readq() / writeq().
> This change propagates consistency in that sense.
> 
> While this is not an issue per se, though if someone wants to address
> it,
> the anchor could be the commit
> 
>   797a796a13df ("asm-generic: architecture independent readq/writeq
> for 32bit environment")
> 
> where non-atomic variants had been introduced.
> 
> Note, there are only few users of above pattern and they will not be
> affected because they do cast returned value. The actual warning has
> been issued on not-yet-upstreamed code.
> 
> Potentially we might get a new warnings if some 64-bit only code
> assigns returned value to unsigned long type of variable. This is
> assumed to be addressed on case-by-case basis.
> 
> Reported-by: lkp <lkp@...el.com>
> Cc: Hitoshi Mitake <mitake@....info.waseda.ac.jp>
> Cc: "Mehta, Sohil" <sohil.mehta@...el.com>
> Signed-off-by: Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
> ---
>  arch/x86/include/asm/io.h | 8 ++++----
>  1 file changed, 4 insertions(+), 4 deletions(-)
> 
> diff --git a/arch/x86/include/asm/io.h b/arch/x86/include/asm/io.h
> index 95e948627fd0..365f5ba9222b 100644
> --- a/arch/x86/include/asm/io.h
> +++ b/arch/x86/include/asm/io.h
> @@ -94,10 +94,10 @@ build_mmio_write(__writel, "l", unsigned int,
> "r", )
>  
>  #ifdef CONFIG_X86_64
>  
> -build_mmio_read(readq, "q", unsigned long, "=r", :"memory")
> -build_mmio_read(__readq, "q", unsigned long, "=r", )
> -build_mmio_write(writeq, "q", unsigned long, "r", :"memory")
> -build_mmio_write(__writeq, "q", unsigned long, "r", )
> +build_mmio_read(readq, "q", unsigned long long, "=r", :"memory")
> +build_mmio_read(__readq, "q", unsigned long long, "=r", )
> +build_mmio_write(writeq, "q", unsigned long long, "r", :"memory")
> +build_mmio_write(__writeq, "q", unsigned long long, "r", )
>  
>  #define readq_relaxed(a)	__readq(a)
>  #define writeq_relaxed(v, a)	__writeq(v, a)

The patch works for me:

Tested-by: Sohil Mehta <sohil.mehta@...el.com>

Sohil