Re: [PATCH] platform/x86/intel/pmt: Sapphire Rapids PMT errata fix

[Hans de Goede <hdegoede@redhat.com>]

Hi,

On 11/5/22 04:42, David E. Box wrote:
> On Sapphire Rapids, due to a hardware issue affecting the PUNIT telemetry
> region, reads that are not done in QWORD quantities and alignment may
> return incorrect data. Use a custom 64-bit copy for this region.
> 
> Signed-off-by: David E. Box <david.e.box@linux.intel.com>

Thank you for your patch, I've applied this patch to my review-hans 
branch:
https://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86.git/log/?h=review-hans

Note it will show up in my review-hans branch once I've pushed my
local branch there, which might take a while.

Once I've run some tests on this branch the patches there will be
added to the platform-drivers-x86/for-next branch and eventually
will be included in the pdx86 pull-request to Linus for the next
merge-window.

Regards,

Hans


> ---
>  drivers/platform/x86/intel/pmt/class.c | 31 +++++++++++++++++++++++++-
>  1 file changed, 30 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/platform/x86/intel/pmt/class.c b/drivers/platform/x86/intel/pmt/class.c
> index 53d7fd2943b4..46598dcb634a 100644
> --- a/drivers/platform/x86/intel/pmt/class.c
> +++ b/drivers/platform/x86/intel/pmt/class.c
> @@ -9,6 +9,7 @@
>   */
>  
>  #include <linux/kernel.h>
> +#include <linux/io-64-nonatomic-lo-hi.h>
>  #include <linux/module.h>
>  #include <linux/mm.h>
>  #include <linux/pci.h>
> @@ -19,6 +20,7 @@
>  #define PMT_XA_START		0
>  #define PMT_XA_MAX		INT_MAX
>  #define PMT_XA_LIMIT		XA_LIMIT(PMT_XA_START, PMT_XA_MAX)
> +#define GUID_SPR_PUNIT		0x9956f43f
>  
>  bool intel_pmt_is_early_client_hw(struct device *dev)
>  {
> @@ -33,6 +35,29 @@ bool intel_pmt_is_early_client_hw(struct device *dev)
>  }
>  EXPORT_SYMBOL_GPL(intel_pmt_is_early_client_hw);
>  
> +static inline int
> +pmt_memcpy64_fromio(void *to, const u64 __iomem *from, size_t count)
> +{
> +	int i, remain;
> +	u64 *buf = to;
> +
> +	if (!IS_ALIGNED((unsigned long)from, 8))
> +		return -EFAULT;
> +
> +	for (i = 0; i < count/8; i++)
> +		buf[i] = readq(&from[i]);
> +
> +	/* Copy any remaining bytes */
> +	remain = count % 8;
> +	if (remain) {
> +		u64 tmp = readq(&from[i]);
> +
> +		memcpy(&buf[i], &tmp, remain);
> +	}
> +
> +	return count;
> +}
> +
>  /*
>   * sysfs
>   */
> @@ -54,7 +79,11 @@ intel_pmt_read(struct file *filp, struct kobject *kobj,
>  	if (count > entry->size - off)
>  		count = entry->size - off;
>  
> -	memcpy_fromio(buf, entry->base + off, count);
> +	if (entry->guid == GUID_SPR_PUNIT)
> +		/* PUNIT on SPR only supports aligned 64-bit read */
> +		count = pmt_memcpy64_fromio(buf, entry->base + off, count);
> +	else
> +		memcpy_fromio(buf, entry->base + off, count);
>  
>  	return count;
>  }
> 
> base-commit: 225469d4acbcb873358d7618bad6e0203b67b964