Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <1c0412a7-7544-9398-c4fa-657b7ea59b94@rivosinc.com>
Date:   Thu, 12 Jan 2023 12:22:10 -0800
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.4.2
Subject: Re: [PATCH] riscv: elf: add .riscv.attributes parsing
Content-Language: en-US
To:     "Eric W. Biederman" <ebiederm@xmission.com>
Cc:     linux-riscv@lists.infradead.org, linux-kernel@vger.kernel.org,
        palmer@rivosinc.com, Paul Walmsley <paul.walmsley@sifive.com>,
        Albert Ou <aou@eecs.berkeley.edu>,
        Kees Cook <keescook@chromium.org>, Guo Ren <guoren@kernel.org>,
        Greentime Hu <greentime.hu@sifive.com>,
        Conor Dooley <conor.dooley@microchip.com>, linux@rivosinc.com,
        kernel test robot <lkp@intel.com>
References: <20230110201841.2069353-1-vineetg@rivosinc.com>
 <87tu0xko6g.fsf@email.froward.int.ebiederm.org>
From:   Vineet Gupta <vineetg@rivosinc.com>
In-Reply-To: <87tu0xko6g.fsf@email.froward.int.ebiederm.org>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk


On 1/11/23 08:53, Eric W. Biederman wrote:
> Vineet Gupta <vineetg@rivosinc.com> writes:
>
>> This implements the elf loader hook to parse RV specific
>> .riscv.attributes section. This section is inserted by compilers
>> (gcc/llvm) with build related information such as -march organized as
>> tag/value attribute pairs.
>>
>> It identifies the various attribute tags (and corresponding values) as
>> currently specified in the psABI specification.
>>
>> This patch only implements the elf parsing mechanics, leaving out the
>> recording/usage of the attributes to subsequent patches.
> Why does the kernel care?
>
> Unless I am mistaken the kind of information you are describing is
> typically dealt with by the dynamic linker, rather than the kernel.
>
> As a general strategy I think it is smart to leave as much to
> userspace and the dynamic linker and possible intead of asking
> the kernel's elf loader to care.

That is true in general. But in some cases we might have to do stuff 
before kick starting usermode such as setting up a shadow stack with 
special permissions etc.

> That allows more rapid
> prototyping and reduces the impact of parsing bugs.