[patch] edac: sb_edac: add sanity check to silence static checker

[patch] edac: sb_edac: add sanity check to silence static checker

[Dan Carpenter]

I assume the the check on if (limit <= prv) prevents n_tads from
actually reaching MAX_TAD.  The problem with that is that it relies
on the hardware returning the right value and Smatch complains that
if it doesn't we could have a buffer overflow.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
---
Feel free to ignore this patch if you want.  I don't have very stong
feelings about this either way.

diff --git a/drivers/edac/sb_edac.c b/drivers/edac/sb_edac.c
index 7a402bf..ebf386c 100644
--- a/drivers/edac/sb_edac.c
+++ b/drivers/edac/sb_edac.c
@@ -970,6 +970,12 @@ static int get_memory_error_data(struct mem_ctl_info *mci,
 			break;
 		prv = limit;
 	}
+	if (n_tads == MAX_TAD) {
+		sprintf(msg, "Could not discover the memory channel");
+		edac_mc_handle_ce_no_info(mci, msg);
+		return -EINVAL;
+	}
+
 	ch_way = TAD_CH(reg) + 1;
 	sck_way = TAD_SOCK(reg) + 1;
 	/*

Re: [patch] edac: sb_edac: add sanity check to silence static checker

[walter harms]

Am 01.11.2011 07:28, schrieb Dan Carpenter:
> I assume the the check on if (limit <= prv) prevents n_tads from
> actually reaching MAX_TAD.  The problem with that is that it relies
> on the hardware returning the right value and Smatch complains that
> if it doesn't we could have a buffer overflow.
> 
> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
> ---
> Feel free to ignore this patch if you want.  I don't have very stong
> feelings about this either way.
> 
> diff --git a/drivers/edac/sb_edac.c b/drivers/edac/sb_edac.c
> index 7a402bf..ebf386c 100644
> --- a/drivers/edac/sb_edac.c
> +++ b/drivers/edac/sb_edac.c
> @@ -970,6 +970,12 @@ static int get_memory_error_data(struct mem_ctl_info *mci,
>  			break;
>  		prv = limit;
>  	}
> +	if (n_tads == MAX_TAD) {
> +		sprintf(msg, "Could not discover the memory channel");

why the sprintf() ? can you not simply:
	edac_mc_handle_ce_no_info(mci,"Could not discover the memory channel");
		
	re,
 	 wh
> +		edac_mc_handle_ce_no_info(mci, msg);
> +		return -EINVAL;
> +	}
> +
>  	ch_way = TAD_CH(reg) + 1;
>  	sck_way = TAD_SOCK(reg) + 1;
>  	/*

Re: [patch] edac: sb_edac: add sanity check to silence static checker

[Mauro Carvalho Chehab]

Em 01-11-2011 10:32, walter harms escreveu:
> 
> 
> Am 01.11.2011 07:28, schrieb Dan Carpenter:
>> I assume the the check on if (limit <= prv) prevents n_tads from
>> actually reaching MAX_TAD.  The problem with that is that it relies
>> on the hardware returning the right value and Smatch complains that
>> if it doesn't we could have a buffer overflow.
>>
>> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
>> ---
>> Feel free to ignore this patch if you want.  I don't have very stong
>> feelings about this either way.
>>
>> diff --git a/drivers/edac/sb_edac.c b/drivers/edac/sb_edac.c
>> index 7a402bf..ebf386c 100644
>> --- a/drivers/edac/sb_edac.c
>> +++ b/drivers/edac/sb_edac.c
>> @@ -970,6 +970,12 @@ static int get_memory_error_data(struct mem_ctl_info *mci,
>>  			break;
>>  		prv = limit;
>>  	}
>> +	if (n_tads == MAX_TAD) {
>> +		sprintf(msg, "Could not discover the memory channel");
> 
> why the sprintf() ? can you not simply:
> 	edac_mc_handle_ce_no_info(mci,"Could not discover the memory channel");

Yes, please us the edac-specific call. I'm working on some patches that will
provide an additional functionality to those edac report calls. So, using
sprintf() won't do the right thing after applying those patches (likely for
Kernel v3.3).

> 		
> 	re,
>  	 wh
>> +		edac_mc_handle_ce_no_info(mci, msg);
>> +		return -EINVAL;
>> +	}
>> +
>>  	ch_way = TAD_CH(reg) + 1;
>>  	sck_way = TAD_SOCK(reg) + 1;
>>  	/*

Re: [patch] edac: sb_edac: add sanity check to silence static checker

[Dan Carpenter]

[-- Attachment #1: Type: text/plain, Size: 1504 bytes --]

On Tue, Nov 01, 2011 at 10:53:18AM -0200, Mauro Carvalho Chehab wrote:
> >> --- a/drivers/edac/sb_edac.c
> >> +++ b/drivers/edac/sb_edac.c
> >> @@ -970,6 +970,12 @@ static int get_memory_error_data(struct mem_ctl_info *mci,
> >>  			break;
> >>  		prv = limit;
> >>  	}
> >> +	if (n_tads == MAX_TAD) {
> >> +		sprintf(msg, "Could not discover the memory channel");
> > 
> > why the sprintf() ? can you not simply:
> > 	edac_mc_handle_ce_no_info(mci,"Could not discover the memory channel");
> 
> Yes, please us the edac-specific call. I'm working on some patches that will
> provide an additional functionality to those edac report calls. So, using
> sprintf() won't do the right thing after applying those patches (likely for
> Kernel v3.3).
> 

I did use the edac-specific call.  Walter is saying that the
sprintf() is not needed because I could just pass the string
directly.  That's true enough, but I'd prefer to leave it how I wrote
it so it matches everything else in this file.  Also passing it
directly as edac_mc_handle_ce_no_info(mci,"Could not discover the memory channel");
makes the line too long for the 80 character limit so someone will
complain if I do that.  This isn't a fast path so the sprintf()
doesn't hurt anything.

Also looking at it now, I suspect this patch might be a bug fix
instead of just to silence the warning.  We have a similar check for
the MAX_SAD loop earlier to the check that I added here for MAX_TAD.

regards,
dan carpenter


[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]