[PATCH] riscv, lib: Fix Zbb strncmp

[PATCH] riscv, lib: Fix Zbb strncmp

[Björn Töpel]

From: Björn Töpel <bjorn@rivosinc.com>

The Zbb optimized strncmp has two parts; a fast path that does XLEN/8B
per iteration, and a slow that does one byte per iteration.

The idea is to compare aligned XLEN chunks for most of strings, and do
the remainder tail in the slow path.

The Zbb strncmp has two issues in the fast path:

Incorrect remainder handling (wrong compare): Assume that the string
length is 9. On 64b systems, the fast path should do one iteration,
and one iteration in the slow path. Instead, both were done in the
fast path, which lead to incorrect results. An example:

  strncmp("/dev/vda", "/dev/", 5);

Correct by changing "bgt" to "bge".

Missing NULL checks in the second string: This could lead to incorrect
results for:

  strncmp("/dev/vda", "/dev/vda\0", 8);

Correct by adding an additional check.

Fixes: b6fcdb191e36 ("RISC-V: add zbb support to string functions")
Suggested-by: Heiko Stuebner <heiko.stuebner@vrull.eu>
Signed-off-by: Björn Töpel <bjorn@rivosinc.com>
---
 arch/riscv/lib/strncmp.S | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/arch/riscv/lib/strncmp.S b/arch/riscv/lib/strncmp.S
index ee49595075be..efd3f3150a54 100644
--- a/arch/riscv/lib/strncmp.S
+++ b/arch/riscv/lib/strncmp.S
@@ -78,11 +78,13 @@ strncmp_zbb:
 	/* Main loop for aligned string.  */
 	.p2align 3
 1:
-	bgt	a0, t6, 3f
+	bge	a0, t6, 3f
 	REG_L	t0, 0(a0)
 	REG_L	t1, 0(a1)
 	orc.b	t3, t0
 	bne	t3, t5, 2f
+	orc.b	t3, t1
+	bne	t3, t5, 2f
 	addi	a0, a0, SZREG
 	addi	a1, a1, SZREG
 	beq	t0, t1, 1b

base-commit: eb9be8310c58c166f9fae3b71c0ad9d6741b4897
-- 
2.37.2


_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv

Re: [PATCH] riscv, lib: Fix Zbb strncmp

[Conor Dooley]

[-- Attachment #1.1: Type: text/plain, Size: 1368 bytes --]

On Tue, Feb 28, 2023 at 07:42:10PM +0100, Björn Töpel wrote:
> From: Björn Töpel <bjorn@rivosinc.com>
> 
> The Zbb optimized strncmp has two parts; a fast path that does XLEN/8B
> per iteration, and a slow that does one byte per iteration.
> 
> The idea is to compare aligned XLEN chunks for most of strings, and do
> the remainder tail in the slow path.
> 
> The Zbb strncmp has two issues in the fast path:
> 
> Incorrect remainder handling (wrong compare): Assume that the string
> length is 9. On 64b systems, the fast path should do one iteration,
> and one iteration in the slow path. Instead, both were done in the
> fast path, which lead to incorrect results. An example:
> 
>   strncmp("/dev/vda", "/dev/", 5);
> 
> Correct by changing "bgt" to "bge".
> 
> Missing NULL checks in the second string: This could lead to incorrect
> results for:
> 
>   strncmp("/dev/vda", "/dev/vda\0", 8);
> 
> Correct by adding an additional check.
> 
> Fixes: b6fcdb191e36 ("RISC-V: add zbb support to string functions")
> Suggested-by: Heiko Stuebner <heiko.stuebner@vrull.eu>
> Signed-off-by: Björn Töpel <bjorn@rivosinc.com>

With your new selftest stuff & on a JH7110:
Tested-by: Conor Dooley <conor.dooley@microchip.com>
Also, given the reporter, this should be:
Reported-by: Guenter Roeck <linux@roeck-us.net>
no?

Thanks,
Conor.

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

[-- Attachment #2: Type: text/plain, Size: 161 bytes --]

_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv

Re: [PATCH] riscv, lib: Fix Zbb strncmp

[Björn Töpel]

Conor Dooley <conor@kernel.org> writes:

> On Tue, Feb 28, 2023 at 07:42:10PM +0100, Björn Töpel wrote:
>> From: Björn Töpel <bjorn@rivosinc.com>
>> 
>> The Zbb optimized strncmp has two parts; a fast path that does XLEN/8B
>> per iteration, and a slow that does one byte per iteration.
>> 
>> The idea is to compare aligned XLEN chunks for most of strings, and do
>> the remainder tail in the slow path.
>> 
>> The Zbb strncmp has two issues in the fast path:
>> 
>> Incorrect remainder handling (wrong compare): Assume that the string
>> length is 9. On 64b systems, the fast path should do one iteration,
>> and one iteration in the slow path. Instead, both were done in the
>> fast path, which lead to incorrect results. An example:
>> 
>>   strncmp("/dev/vda", "/dev/", 5);
>> 
>> Correct by changing "bgt" to "bge".
>> 
>> Missing NULL checks in the second string: This could lead to incorrect
>> results for:
>> 
>>   strncmp("/dev/vda", "/dev/vda\0", 8);
>> 
>> Correct by adding an additional check.
>> 
>> Fixes: b6fcdb191e36 ("RISC-V: add zbb support to string functions")
>> Suggested-by: Heiko Stuebner <heiko.stuebner@vrull.eu>
>> Signed-off-by: Björn Töpel <bjorn@rivosinc.com>
>
> With your new selftest stuff & on a JH7110:
> Tested-by: Conor Dooley <conor.dooley@microchip.com>
> Also, given the reporter, this should be:
> Reported-by: Guenter Roeck <linux@roeck-us.net>
> no?

Indeed. Sorry, Guenter, for missing that!


_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv

Re: [PATCH] riscv, lib: Fix Zbb strncmp

[Guenter Roeck]

On 2/28/23 11:12, Conor Dooley wrote:
> On Tue, Feb 28, 2023 at 07:42:10PM +0100, Björn Töpel wrote:
>> From: Björn Töpel <bjorn@rivosinc.com>
>>
>> The Zbb optimized strncmp has two parts; a fast path that does XLEN/8B
>> per iteration, and a slow that does one byte per iteration.
>>
>> The idea is to compare aligned XLEN chunks for most of strings, and do
>> the remainder tail in the slow path.
>>
>> The Zbb strncmp has two issues in the fast path:
>>
>> Incorrect remainder handling (wrong compare): Assume that the string
>> length is 9. On 64b systems, the fast path should do one iteration,
>> and one iteration in the slow path. Instead, both were done in the
>> fast path, which lead to incorrect results. An example:
>>
>>    strncmp("/dev/vda", "/dev/", 5);
>>
>> Correct by changing "bgt" to "bge".
>>
>> Missing NULL checks in the second string: This could lead to incorrect
>> results for:
>>
>>    strncmp("/dev/vda", "/dev/vda\0", 8);
>>
>> Correct by adding an additional check.
>>
>> Fixes: b6fcdb191e36 ("RISC-V: add zbb support to string functions")
>> Suggested-by: Heiko Stuebner <heiko.stuebner@vrull.eu>
>> Signed-off-by: Björn Töpel <bjorn@rivosinc.com>
> 
> With your new selftest stuff & on a JH7110:
> Tested-by: Conor Dooley <conor.dooley@microchip.com>
> Also, given the reporter, this should be:
> Reported-by: Guenter Roeck <linux@roeck-us.net>
> no?
> 

No worries. I'll give the patch a try.

Guenter



_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv

Re: [PATCH] riscv, lib: Fix Zbb strncmp

[Guenter Roeck]

On 2/28/23 10:42, Björn Töpel wrote:
> From: Björn Töpel <bjorn@rivosinc.com>
> 
> The Zbb optimized strncmp has two parts; a fast path that does XLEN/8B
> per iteration, and a slow that does one byte per iteration.
> 
> The idea is to compare aligned XLEN chunks for most of strings, and do
> the remainder tail in the slow path.
> 
> The Zbb strncmp has two issues in the fast path:
> 
> Incorrect remainder handling (wrong compare): Assume that the string
> length is 9. On 64b systems, the fast path should do one iteration,
> and one iteration in the slow path. Instead, both were done in the
> fast path, which lead to incorrect results. An example:
> 
>    strncmp("/dev/vda", "/dev/", 5);
> 
> Correct by changing "bgt" to "bge".
> 
> Missing NULL checks in the second string: This could lead to incorrect
> results for:
> 
>    strncmp("/dev/vda", "/dev/vda\0", 8);
> 
> Correct by adding an additional check.
> 
> Fixes: b6fcdb191e36 ("RISC-V: add zbb support to string functions")
> Suggested-by: Heiko Stuebner <heiko.stuebner@vrull.eu>
> Signed-off-by: Björn Töpel <bjorn@rivosinc.com>

Tested-by: Guenter Roeck <linux@roeck-us.net>

> ---
>   arch/riscv/lib/strncmp.S | 4 +++-
>   1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/riscv/lib/strncmp.S b/arch/riscv/lib/strncmp.S
> index ee49595075be..efd3f3150a54 100644
> --- a/arch/riscv/lib/strncmp.S
> +++ b/arch/riscv/lib/strncmp.S
> @@ -78,11 +78,13 @@ strncmp_zbb:
>   	/* Main loop for aligned string.  */
>   	.p2align 3
>   1:
> -	bgt	a0, t6, 3f
> +	bge	a0, t6, 3f
>   	REG_L	t0, 0(a0)
>   	REG_L	t1, 0(a1)
>   	orc.b	t3, t0
>   	bne	t3, t5, 2f
> +	orc.b	t3, t1
> +	bne	t3, t5, 2f
>   	addi	a0, a0, SZREG
>   	addi	a1, a1, SZREG
>   	beq	t0, t1, 1b
> 
> base-commit: eb9be8310c58c166f9fae3b71c0ad9d6741b4897


_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv

Re: [PATCH] riscv, lib: Fix Zbb strncmp

[Palmer Dabbelt]

On Tue, 28 Feb 2023 13:18:12 PST (-0800), linux@roeck-us.net wrote:
> On 2/28/23 10:42, Björn Töpel wrote:
>> From: Björn Töpel <bjorn@rivosinc.com>
>>
>> The Zbb optimized strncmp has two parts; a fast path that does XLEN/8B
>> per iteration, and a slow that does one byte per iteration.
>>
>> The idea is to compare aligned XLEN chunks for most of strings, and do
>> the remainder tail in the slow path.
>>
>> The Zbb strncmp has two issues in the fast path:
>>
>> Incorrect remainder handling (wrong compare): Assume that the string
>> length is 9. On 64b systems, the fast path should do one iteration,
>> and one iteration in the slow path. Instead, both were done in the
>> fast path, which lead to incorrect results. An example:
>>
>>    strncmp("/dev/vda", "/dev/", 5);
>>
>> Correct by changing "bgt" to "bge".
>>
>> Missing NULL checks in the second string: This could lead to incorrect
>> results for:
>>
>>    strncmp("/dev/vda", "/dev/vda\0", 8);
>>
>> Correct by adding an additional check.
>>
>> Fixes: b6fcdb191e36 ("RISC-V: add zbb support to string functions")
>> Suggested-by: Heiko Stuebner <heiko.stuebner@vrull.eu>
>> Signed-off-by: Björn Töpel <bjorn@rivosinc.com>
>
> Tested-by: Guenter Roeck <linux@roeck-us.net>

Thanks guys, I'm going to apply this ASAP as it's a pretty concrete 
breakage in new code from this merge window.  I just stuck it in my 
staging branch, it should show up on for-next after the next batch of 
tests.

>
>> ---
>>   arch/riscv/lib/strncmp.S | 4 +++-
>>   1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/arch/riscv/lib/strncmp.S b/arch/riscv/lib/strncmp.S
>> index ee49595075be..efd3f3150a54 100644
>> --- a/arch/riscv/lib/strncmp.S
>> +++ b/arch/riscv/lib/strncmp.S
>> @@ -78,11 +78,13 @@ strncmp_zbb:
>>   	/* Main loop for aligned string.  */
>>   	.p2align 3
>>   1:
>> -	bgt	a0, t6, 3f
>> +	bge	a0, t6, 3f
>>   	REG_L	t0, 0(a0)
>>   	REG_L	t1, 0(a1)
>>   	orc.b	t3, t0
>>   	bne	t3, t5, 2f
>> +	orc.b	t3, t1
>> +	bne	t3, t5, 2f
>>   	addi	a0, a0, SZREG
>>   	addi	a1, a1, SZREG
>>   	beq	t0, t1, 1b
>>
>> base-commit: eb9be8310c58c166f9fae3b71c0ad9d6741b4897

_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv

Re: [PATCH] riscv, lib: Fix Zbb strncmp

[patchwork-bot+linux-riscv]

Hello:

This patch was applied to riscv/linux.git (for-next)
by Palmer Dabbelt <palmer@rivosinc.com>:

On Tue, 28 Feb 2023 19:42:10 +0100 you wrote:
> From: Björn Töpel <bjorn@rivosinc.com>
> 
> The Zbb optimized strncmp has two parts; a fast path that does XLEN/8B
> per iteration, and a slow that does one byte per iteration.
> 
> The idea is to compare aligned XLEN chunks for most of strings, and do
> the remainder tail in the slow path.
> 
> [...]

Here is the summary with links:
  - riscv, lib: Fix Zbb strncmp
    https://git.kernel.org/riscv/c/81a1dd10b072

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html



_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv