Re: [PATCH v4 1/2] efi/cper: add cper_mem_err_status_str to decode error description

[Shuai Xue <xueshuai@linux.alibaba.com>]

Hi, Borislav,

Thank you for your comments.

在 2022/1/26 AM3:08, Borislav Petkov 写道:
> On Tue, Jan 25, 2022 at 10:49:38AM +0800, Shuai Xue wrote:
>> Introduce a new helper function cper_mem_err_status_str() which is used to
>> decode the description of error status, and the cper_print_mem() will call
>> it and report the details of error status.
>>
>> Signed-off-by: Shuai Xue <xueshuai@linux.alibaba.com>
>> ---
>>  drivers/firmware/efi/cper.c | 29 ++++++++++++++++++++++++++++-
>>  include/linux/cper.h        |  1 +
>>  2 files changed, 29 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/firmware/efi/cper.c b/drivers/firmware/efi/cper.c
>> index 6ec8edec6329..7f08d4ea906e 100644
>> --- a/drivers/firmware/efi/cper.c
>> +++ b/drivers/firmware/efi/cper.c
>> @@ -211,6 +211,31 @@ const char *cper_mem_err_type_str(unsigned int etype)
>>  }
>>  EXPORT_SYMBOL_GPL(cper_mem_err_type_str);
>>  
>> +const char *cper_mem_err_status_str(u64 status)
>> +{
>> +	switch ((status >> 8) & 0xff) {
>> +	case  1:	return "Error detected internal to the component";
>> +	case  4:	return "Storage error in DRAM memory";
>> +	case  5:	return "Storage error in TLB";
>> +	case  6:	return "Storage error in cache";
>> +	case  7:	return "Error in one or more functional units";
>> +	case  8:	return "component failed self test";
> 
> Well, at least start them all with capital letters: "Component... " And
> yes, I know this is how it is in the spec but the spec has typos and
> other problems - doesn't mean we have to copy them too.

You are right, I will fix it in next version.

>> +	case  9:	return "Overflow or undervalue of internal queue";
>> +	case 16:	return "Error detected in the bus";
>> +	case 17:	return "Virtual address not found on IO-TLB or IO-PDIR";
>> +	case 18:	return "Improper access error";
>> +	case 19:	return "Access to a memory address which is not mapped to any component";
>> +	case 20:	return "Loss of Lockstep";
>> +	case 21:	return "Response not associated with a request";
>> +	case 22:	return "Bus parity error - must also set the A, C, or D Bits";
>> +	case 23:	return "Detection of a PATH_ERROR ";
> 
> Trailing space here.

Sorry, will delete it.

> Also what is PATH_ERROR?
> 
> That "PATH_ERROR" is nowhere else explained in that big fat UEFI spec.
> 2558 pages and they can't explain *that*. Geez.

I don't know either. A related item I found is "iSCSI Device Path error".
Section 10 defines the device path protocol, but I don't know if "PATH_ERROR"
means the path of the device is not found, or something else.

>> +	case 25:	return "Bus operation timeout";
>> +	case 26:	return "A read was issued to data that has been poisoned";
>> +	default:	return "reserved";
>> +	}
>> +}
>> +EXPORT_SYMBOL_GPL(cper_mem_err_status_str);
>> +
>>  static int cper_mem_err_location(struct cper_mem_err_compact *mem, char *msg)
>>  {
>>  	u32 len, n;
>> @@ -334,7 +359,9 @@ static void cper_print_mem(const char *pfx, const struct cper_sec_mem_err *mem,
>>  		return;
>>  	}
>>  	if (mem->validation_bits & CPER_MEM_VALID_ERROR_STATUS)
>> -		printk("%s""error_status: 0x%016llx\n", pfx, mem->error_status);
>> +		printk("%s""error_status: %s (0x%016llx)\n",
> 
> Why do you insist on having two back-to-back strings instead of one
> here?
> 
> (And don't tell me it is because the other function calls here do it
> too.)
> 
> FWIW, even checkpatch complains here:
> 
> WARNING: Consecutive strings are generally better as a single string
> #87: FILE: drivers/firmware/efi/cper.c:362:
> +               printk("%s""error_status: %s (0x%016llx)\n",
> 
> Btw, please integrate scripts/checkpatch.pl into your patch creation
> workflow. Some of the warnings/errors *actually* make sense.

Sorry, I did integrate scripts/checkpatch.pl before sending patch. And as you see,
other function calls here do the same, so I ignored the warnings. I will change as
your comments in next version:

-		printk("%s""error_status: 0x%016llx\n", pfx, mem->error_status);
+		printk("%s error_status: %s (0x%016llx)\n",


Best regards,
Shuai