Re: [PATCH] x86/mce: /dev/mcelog: Dynamically allocate space for machine check records

[Borislav Petkov]

On Fri, Feb 07, 2020 at 04:05:51PM -0800, Tony Luck wrote:
> We have had a hard coded limit of 32 machine check records since the
> dawn of time.  But as numbers of cores increase, it is possible for
> more than 32 errors to be reported before a user process reads from
> /dev/mcelog. In this case the additional errors are lost.
> 
> Keep 32 as the minimum. But tune the maximum value up based on the
> number of processors.
> 
> Signed-off-by: Tony Luck <tony.luck@xxxxxxxxx>
> ---
>  arch/x86/include/asm/mce.h           |  6 ++--
>  arch/x86/kernel/cpu/mce/dev-mcelog.c | 46 ++++++++++++++++------------
>  2 files changed, 29 insertions(+), 23 deletions(-)

...

> @@ -214,21 +210,21 @@ static ssize_t mce_chrdev_read(struct file *filp, char __user *ubuf,
>  
>  	/* Only supports full reads right now */
>  	err = -EINVAL;
> -	if (*off != 0 || usize < MCE_LOG_LEN*sizeof(struct mce))
> +	if (*off != 0 || usize < mcelog->len*sizeof(struct mce))

Add spaces around that *

>  		goto out;
>  
> -	next = mcelog.next;
> +	next = mcelog->next;
>  	err = 0;
>  
>  	for (i = 0; i < next; i++) {
> -		struct mce *m = &mcelog.entry[i];
> +		struct mce *m = &mcelog->entry[i];
>  
>  		err |= copy_to_user(buf, m, sizeof(*m));
>  		buf += sizeof(*m);
>  	}
>  
> -	memset(mcelog.entry, 0, next * sizeof(struct mce));
> -	mcelog.next = 0;
> +	memset(mcelog->entry, 0, next * sizeof(struct mce));
> +	mcelog->next = 0;
>  
>  	if (err)
>  		err = -EFAULT;

...

> @@ -340,6 +336,15 @@ static struct miscdevice mce_chrdev_device = {
>  static __init int dev_mcelog_init_device(void)
>  {
>  	int err;
> +	int mce_log_len;

Please sort function local variables declaration in a reverse christmas
tree order:

	<type A> longest_variable_name;
	<type B> shorter_var_name;
	<type C> even_shorter;
	<type D> i;

> +
> +	mce_log_len = max(MCE_LOG_MIN_LEN, num_online_cpus());

arch/x86/kernel/cpu/mce/dev-mcelog.c: In function âdev_mcelog_init_deviceâ:
./include/linux/kernel.h:835:29: warning: comparison of distinct pointer types lacks a cast
  835 |   (!!(sizeof((typeof(x) *)1 == (typeof(y) *)1)))
      |                             ^~
./include/linux/kernel.h:849:4: note: in expansion of macro â__typecheckâ
  849 |   (__typecheck(x, y) && __no_side_effects(x, y))
      |    ^~~~~~~~~~~
./include/linux/kernel.h:859:24: note: in expansion of macro â__safe_cmpâ
  859 |  __builtin_choose_expr(__safe_cmp(x, y), \
      |                        ^~~~~~~~~~
./include/linux/kernel.h:875:19: note: in expansion of macro â__careful_cmpâ
  875 | #define max(x, y) __careful_cmp(x, y, >)
      |                   ^~~~~~~~~~~~~
arch/x86/kernel/cpu/mce/dev-mcelog.c:341:16: note: in expansion of macro âmaxâ
  341 |  mce_log_len = max(MCE_LOG_MIN_LEN, num_online_cpus());
      |                ^~~

That MCE_LOG_MIN_LEN wants to be 32U.

> +	mcelog = kzalloc(sizeof(*mcelog) + mce_log_len * sizeof(struct mce), GFP_KERNEL);
> +	if (!mcelog)
> +		return -ENOMEM;

<---- newline here.

> +	strncpy(mcelog->signature, MCE_LOG_SIGNATURE, sizeof(mcelog->signature));
> +	mcelog->len = mce_log_len;
> +	mcelog->recordlen = sizeof(struct mce);
>  
>  	/* register character device /dev/mcelog */
>  	err = misc_register(&mce_chrdev_device);

Thx.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette