Received-SPF: pass (google.com: domain of linux-kernel+bounces-147379-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
Message-ID: <ece31eca-37bc-4b39-9fc9-7e6fda741729@zytor.com>
Date: Tue, 16 Apr 2024 11:23:24 -0700
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [PATCH v2 1/1] x86/fred: Fix INT80 emulation for FRED
Content-Language: en-US
To: "Xin Li (Intel)" <xin@zytor.com>, linux-kernel@vger.kernel.org
Cc: luto@kernel.org, tglx@linutronix.de, mingo@redhat.com, bp@alien8.de,
        dave.hansen@linux.intel.com, x86@kernel.org
References: <20240416175804.3328999-1-xin@zytor.com>
From: "H. Peter Anvin" <hpa@zytor.com>
In-Reply-To: <20240416175804.3328999-1-xin@zytor.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit

On 4/16/24 10:58, Xin Li (Intel) wrote:
> Add a FRED-specific INT80 handler fred_int80_emulation().
> 
> Commit
>    55617fb991df ("x86/entry: Do not allow external 0x80 interrupts")
> added a bunch of checks to the int $0x80 path, however they are
> unnecessary and event wrong in fact under FRED.


I think the following points should be added to the head comment, and 
not just the commit log:

> 1) FRED distinguishes external interrupts from software interrupts,
>     thus int80_emulation() should NEVER be called for handling an
>     external interrupt, and then int80_is_external() is meaningless
>     when FRED is enabled.

1a) As INT instructions and hardware interrupts are separate event 
types, FRED does not preclude the use of vector 0x80 for external 
interrupts. As a result the FRED setup code does *NOT* reserve vector 
0x80 and calling int80_is_external() is not merely suboptimal but 
actively incorrect: it could could cause a system call to be incorrectly 
ignored.

> 2) The FRED kernel entry handler NEVER dispatches INTx, which is
>     of event type EVENT_TYPE_SWINT, so the user mode checking in
>     do_int80_emulation() is redundant.
> 
> 3) int80_emulation() does a CLEAR_BRANCH_HISTORY, which is likly
s/likly/likely/
>     an overkill for new x86 CPU implementations that support FRED.
s/an //

4) int $0x80 is the FAST path for 32-bit system calls under FRED.

[...]

> A dedicated FRED INT80 handler duplicates most of the code in
s/most/quite a bit/

(I think there is actually less than half the code left. This could be 
further cleaned up by inlining the common code, but if I were still 
maintainer I would not want that for x86/urgent. This patch has the very 
nice property for x86/urgent purposes that it doesn't touch non-FRED 
code at all.)

> do_int80_emulation(), but it avoids sprinkling moar tests and
s/moar/more/
> seems more readable.
> 
> Suggested-by: H. Peter Anvin (Intel) <hpa@zytor.com>
> Signed-off-by: Xin Li (Intel) <xin@zytor.com>

So, as Borislav pointed out:

Fixes: 55617fb991df ("x86/entry: Do not allow external 0x80 interrupts")

My fault.