[PATCH v1] riscv: cpu: improve multi-letter extension detection in supports_extension()

[PATCH v1] riscv: cpu: improve multi-letter extension detection in supports_extension()

[Conor Dooley]

From: Conor Dooley <conor.dooley@microchip.com>

The first multi-letter extension after the single-letter extensions does
not have to be preceded by an underscore, which could cause the parser
to mistakenly find a single-letter extension after the start of the
multi-letter portion of the string.
Three letters precede multi-letter extensions (s, x & z), none of which
are valid single-letter extensions. The dt-binding also allows
multi-letter extensions starting with h, but no such extension have been
frozen or ratified, and the unprivileged spec no longer uses "h" as a
prefix for multi-letter hypervisor extensions, having moved to "sh"
instead. For that reason, modify the parser to stop at s, x & z to prevent
this overrun, ignoring h.

Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
---
The parser in U-Boot only supports single-letter extensions & the
single-letter h has to be at the end of the single-letter section, so it
would not be difficult to terminate parsing once a h is seen (you'd need
to support the hypervisor extension to support additional hypervisor
extensions after all) if in the future a multi-letter extension starting
with h did come about. I've got no problem adding a special case for h,
but I'm tempted to just remove the multi-letter h extensions from the
binding, given there's actually not going to be any extensions ratified
using that naming scheme.

CC: Rick Chen <rick@andestech.com>
CC: Leo <ycliang@andestech.com>
CC: Tom Rini <trini@konsulko.com>
CC: Simon Glass <sjg@chromium.org>
CC: Chanho Park <chanho61.park@samsung.com>
CC: Heinrich Schuchardt <xypron.glpk@gmx.de>
CC: Bin Meng <bmeng.cn@gmail.com>
CC: Conor Dooley <conor.dooley@microchip.com>
CC: palmer@dabbelt.com
CC: u-boot@lists.denx.de
---
 arch/riscv/cpu/cpu.c | 22 ++++++++++++++++------
 1 file changed, 16 insertions(+), 6 deletions(-)

diff --git a/arch/riscv/cpu/cpu.c b/arch/riscv/cpu/cpu.c
index 8445c5823e..ecfefa1a02 100644
--- a/arch/riscv/cpu/cpu.c
+++ b/arch/riscv/cpu/cpu.c
@@ -49,14 +49,24 @@ static inline bool supports_extension(char ext)
 	}
 	if (!cpu_get_desc(dev, desc, sizeof(desc))) {
 		/*
-		 * skip the first 4 characters (rv32|rv64) and
-		 * check until underscore
+		 * skip the first 4 characters (rv32|rv64)
 		 */
 		for (i = 4; i < sizeof(desc); i++) {
-			if (desc[i] == '_' || desc[i] == '\0')
-				break;
-			if (desc[i] == ext)
-				return true;
+			switch (desc[i]) {
+			case 's':
+			case 'x':
+			case 'z':
+			case '_':
+			case '\0':
+				/*
+				 * Any of these characters mean the single
+				 * letter extensions have all been consumed.
+				 */
+				return false;
+			default:
+				if (desc[i] == ext)
+					return true;
+			}
 		}
 	}
 
-- 
2.43.0

Re: [PATCH v1] riscv: cpu: improve multi-letter extension detection in supports_extension()

[Heinrich Schuchardt]

On 3/5/24 00:28, Conor Dooley wrote:
> From: Conor Dooley <conor.dooley@microchip.com>
>
> The first multi-letter extension after the single-letter extensions does
> not have to be preceded by an underscore, which could cause the parser
> to mistakenly find a single-letter extension after the start of the
> multi-letter portion of the string.
> Three letters precede multi-letter extensions (s, x & z), none of which
> are valid single-letter extensions. The dt-binding also allows
> multi-letter extensions starting with h, but no such extension have been
> frozen or ratified, and the unprivileged spec no longer uses "h" as a
> prefix for multi-letter hypervisor extensions, having moved to "sh"
> instead. For that reason, modify the parser to stop at s, x & z to prevent
> this overrun, ignoring h.
>
> Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
> ---
> The parser in U-Boot only supports single-letter extensions & the
> single-letter h has to be at the end of the single-letter section, so it
> would not be difficult to terminate parsing once a h is seen (you'd need
> to support the hypervisor extension to support additional hypervisor
> extensions after all) if in the future a multi-letter extension starting
> with h did come about. I've got no problem adding a special case for h,
> but I'm tempted to just remove the multi-letter h extensions from the
> binding, given there's actually not going to be any extensions ratified
> using that naming scheme.
>
> CC: Rick Chen <rick@andestech.com>
> CC: Leo <ycliang@andestech.com>
> CC: Tom Rini <trini@konsulko.com>
> CC: Simon Glass <sjg@chromium.org>
> CC: Chanho Park <chanho61.park@samsung.com>
> CC: Heinrich Schuchardt <xypron.glpk@gmx.de>
> CC: Bin Meng <bmeng.cn@gmail.com>
> CC: Conor Dooley <conor.dooley@microchip.com>
> CC: palmer@dabbelt.com
> CC: u-boot@lists.denx.de
> ---
>   arch/riscv/cpu/cpu.c | 22 ++++++++++++++++------
>   1 file changed, 16 insertions(+), 6 deletions(-)
>
> diff --git a/arch/riscv/cpu/cpu.c b/arch/riscv/cpu/cpu.c
> index 8445c5823e..ecfefa1a02 100644
> --- a/arch/riscv/cpu/cpu.c
> +++ b/arch/riscv/cpu/cpu.c
> @@ -49,14 +49,24 @@ static inline bool supports_extension(char ext)
>   	}
>   	if (!cpu_get_desc(dev, desc, sizeof(desc))) {
>   		/*
> -		 * skip the first 4 characters (rv32|rv64) and
> -		 * check until underscore
> +		 * skip the first 4 characters (rv32|rv64)
>   		 */
>   		for (i = 4; i < sizeof(desc); i++) {
> -			if (desc[i] == '_' || desc[i] == '\0')
> -				break;
> -			if (desc[i] == ext)
> -				return true;
> +			switch (desc[i]) {
> +			case 's':
> +			case 'x':
> +			case 'z':
> +			case '_':
> +			case '\0':
> +				/*
> +				 * Any of these characters mean the single
> +				 * letter extensions have all been consumed.
> +				 */
> +				return false;
> +			default:
> +				if (desc[i] == ext)
> +					return true;
> +			}
>   		}
>   	}
>

According to
https://github.com/riscv/riscv-isa-manual/blob/main/src/naming.adoc the
ISA string is case insensitive. Why can we assume here that it is lower
case?

Best regards

Heinrich

Re: [PATCH v1] riscv: cpu: improve multi-letter extension detection in supports_extension()

[Conor Dooley]

[-- Attachment #1: Type: text/plain, Size: 371 bytes --]

On Tue, Mar 05, 2024 at 08:34:20AM +0100, Heinrich Schuchardt wrote:
> On 3/5/24 00:28, Conor Dooley wrote:
> > From: Conor Dooley <conor.dooley@microchip.com>
> According to
> https://github.com/riscv/riscv-isa-manual/blob/main/src/naming.adoc the
> ISA string is case insensitive. Why can we assume here that it is lower
> case?

The binding does not allow upper-case.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

Re: [PATCH v1] riscv: cpu: improve multi-letter extension detection in supports_extension()

[Heinrich Schuchardt]

On 3/5/24 08:54, Conor Dooley wrote:
> On Tue, Mar 05, 2024 at 08:34:20AM +0100, Heinrich Schuchardt wrote:
>> On 3/5/24 00:28, Conor Dooley wrote:
>>> From: Conor Dooley <conor.dooley@microchip.com>
>> According to
>> https://github.com/riscv/riscv-isa-manual/blob/main/src/naming.adoc the
>> ISA string is case insensitive. Why can we assume here that it is lower
>> case?
> 
> The binding does not allow upper-case.

Ok. That is in Linux' 
Documentation/devicetree/bindings/riscv/extensions.yaml.

Reviewed-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>

Re: [PATCH v1] riscv: cpu: improve multi-letter extension detection in supports_extension()

[Conor Dooley]

[-- Attachment #1: Type: text/plain, Size: 660 bytes --]

On Tue, Mar 05, 2024 at 09:10:59AM +0100, Heinrich Schuchardt wrote:
> On 3/5/24 08:54, Conor Dooley wrote:
> > On Tue, Mar 05, 2024 at 08:34:20AM +0100, Heinrich Schuchardt wrote:
> > > On 3/5/24 00:28, Conor Dooley wrote:
> > > > From: Conor Dooley <conor.dooley@microchip.com>
> > > According to
> > > https://github.com/riscv/riscv-isa-manual/blob/main/src/naming.adoc the
> > > ISA string is case insensitive. Why can we assume here that it is lower
> > > case?
> > 
> > The binding does not allow upper-case.
> 
> Ok. That is in Linux'
> Documentation/devicetree/bindings/riscv/extensions.yaml.

Right. I should've linked to that, sorry.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]