[PATCH] [RFC] usb: gadget: avoid unusual inline assembly

[PATCH] [RFC] usb: gadget: avoid unusual inline assembly

[Arnd Bergmann]

From: Arnd Bergmann <arnd@arndb.de>

clang does not understand the "mrc%?" syntax:

drivers/usb/gadget/udc/pxa25x_udc.c:2330:11: error: invalid % escape in inline assembly string

I don't understand it either, but removing the %? here gets it to build.
This is probably wrong and someone else should do a proper patch.

Any suggestions?

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
 drivers/usb/gadget/udc/pxa25x_udc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/usb/gadget/udc/pxa25x_udc.c b/drivers/usb/gadget/udc/pxa25x_udc.c
index a09ec1d826b2..52cdfd8212d6 100644
--- a/drivers/usb/gadget/udc/pxa25x_udc.c
+++ b/drivers/usb/gadget/udc/pxa25x_udc.c
@@ -2325,7 +2325,7 @@ static int pxa25x_udc_probe(struct platform_device *pdev)
 	pr_info("%s: version %s\n", driver_name, DRIVER_VERSION);
 
 	/* insist on Intel/ARM/XScale */
-	asm("mrc%? p15, 0, %0, c0, c0" : "=r" (chiprev));
+	asm("mrc p15, 0, %0, c0, c0" : "=r" (chiprev));
 	if ((chiprev & CP15R0_VENDOR_MASK) != CP15R0_XSCALE_VALUE) {
 		pr_err("%s: not XScale!\n", driver_name);
 		return -ENODEV;
-- 
2.29.2

Re: [PATCH] [RFC] usb: gadget: avoid unusual inline assembly

[Greg Kroah-Hartman]

On Mon, Sep 27, 2021 at 02:38:20PM +0200, Arnd Bergmann wrote:
> From: Arnd Bergmann <arnd@arndb.de>
> 
> clang does not understand the "mrc%?" syntax:
> 
> drivers/usb/gadget/udc/pxa25x_udc.c:2330:11: error: invalid % escape in inline assembly string
> 
> I don't understand it either, but removing the %? here gets it to build.
> This is probably wrong and someone else should do a proper patch.
> 
> Any suggestions?
> 
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> ---
>  drivers/usb/gadget/udc/pxa25x_udc.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/usb/gadget/udc/pxa25x_udc.c b/drivers/usb/gadget/udc/pxa25x_udc.c
> index a09ec1d826b2..52cdfd8212d6 100644
> --- a/drivers/usb/gadget/udc/pxa25x_udc.c
> +++ b/drivers/usb/gadget/udc/pxa25x_udc.c
> @@ -2325,7 +2325,7 @@ static int pxa25x_udc_probe(struct platform_device *pdev)
>  	pr_info("%s: version %s\n", driver_name, DRIVER_VERSION);
>  
>  	/* insist on Intel/ARM/XScale */
> -	asm("mrc%? p15, 0, %0, c0, c0" : "=r" (chiprev));
> +	asm("mrc p15, 0, %0, c0, c0" : "=r" (chiprev));
>  	if ((chiprev & CP15R0_VENDOR_MASK) != CP15R0_XSCALE_VALUE) {
>  		pr_err("%s: not XScale!\n", driver_name);
>  		return -ENODEV;
> -- 
> 2.29.2
> 

Given that no one had any objections, I'll queue this up and see what
breaks :)

thanks,

greg k-h

Re: [PATCH] [RFC] usb: gadget: avoid unusual inline assembly

[Nick Desaulniers]

On Mon, Sep 27, 2021 at 5:38 AM Arnd Bergmann <arnd@kernel.org> wrote:
>
> From: Arnd Bergmann <arnd@arndb.de>
>
> clang does not understand the "mrc%?" syntax:
>
> drivers/usb/gadget/udc/pxa25x_udc.c:2330:11: error: invalid % escape in inline assembly string
>
> I don't understand it either, but removing the %? here gets it to build.
> This is probably wrong and someone else should do a proper patch.
>
> Any suggestions?

https://gcc.gnu.org/onlinedocs/gccint/Output-Template.html says:
>> ‘%’ followed by a punctuation character specifies a substitution that does not use an operand. Only one case is standard: ‘%%’ outputs a ‘%’ into the assembler code.

I don't think the intention was to have `mrc?` in the emitted
assembler (is that even valid?), so it's not clear to me what this
point of `%?` was. Patch LGTM.

>
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> ---
>  drivers/usb/gadget/udc/pxa25x_udc.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/usb/gadget/udc/pxa25x_udc.c b/drivers/usb/gadget/udc/pxa25x_udc.c
> index a09ec1d826b2..52cdfd8212d6 100644
> --- a/drivers/usb/gadget/udc/pxa25x_udc.c
> +++ b/drivers/usb/gadget/udc/pxa25x_udc.c
> @@ -2325,7 +2325,7 @@ static int pxa25x_udc_probe(struct platform_device *pdev)
>         pr_info("%s: version %s\n", driver_name, DRIVER_VERSION);
>
>         /* insist on Intel/ARM/XScale */
> -       asm("mrc%? p15, 0, %0, c0, c0" : "=r" (chiprev));
> +       asm("mrc p15, 0, %0, c0, c0" : "=r" (chiprev));
>         if ((chiprev & CP15R0_VENDOR_MASK) != CP15R0_XSCALE_VALUE) {
>                 pr_err("%s: not XScale!\n", driver_name);
>                 return -ENODEV;
> --
> 2.29.2
>


-- 
Thanks,
~Nick Desaulniers

Re: [PATCH] [RFC] usb: gadget: avoid unusual inline assembly

[Florian Fainelli]

On 9/27/21 5:38 AM, Arnd Bergmann wrote:
> From: Arnd Bergmann <arnd@arndb.de>
> 
> clang does not understand the "mrc%?" syntax:
> 
> drivers/usb/gadget/udc/pxa25x_udc.c:2330:11: error: invalid % escape in inline assembly string
> 
> I don't understand it either, but removing the %? here gets it to build.
> This is probably wrong and someone else should do a proper patch.
> 
> Any suggestions?
> 
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> ---
>  drivers/usb/gadget/udc/pxa25x_udc.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/usb/gadget/udc/pxa25x_udc.c b/drivers/usb/gadget/udc/pxa25x_udc.c
> index a09ec1d826b2..52cdfd8212d6 100644
> --- a/drivers/usb/gadget/udc/pxa25x_udc.c
> +++ b/drivers/usb/gadget/udc/pxa25x_udc.c
> @@ -2325,7 +2325,7 @@ static int pxa25x_udc_probe(struct platform_device *pdev)
>  	pr_info("%s: version %s\n", driver_name, DRIVER_VERSION);
>  
>  	/* insist on Intel/ARM/XScale */
> -	asm("mrc%? p15, 0, %0, c0, c0" : "=r" (chiprev));
> +	asm("mrc p15, 0, %0, c0, c0" : "=r" (chiprev));

You could consider using read_cpuid() from
arch/arm/include/asm/cputype.h as this driver really does not seem to
have portability in mind.

>  	if ((chiprev & CP15R0_VENDOR_MASK) != CP15R0_XSCALE_VALUE) {
>  		pr_err("%s: not XScale!\n", driver_name);
>  		return -ENODEV;
> 


-- 
Florian