Re: [PATCH printk v3 03/14] printk: Use prb_first_seq() as base for 32bit seq macros

[Petr Mladek <pmladek@suse.com>]

On Thu 2023-12-14 22:47:50, John Ogness wrote:
> Note: This change only applies to 32bit architectures. On 64bit
>       architectures the macros are NOPs.
> 
> Currently prb_next_seq() is used as the base for the 32bit seq
> macros __u64seq_to_ulseq() and __ulseq_to_u64seq(). However, in
> a follow-up commit, prb_next_seq() will need to make use of the
> 32bit seq macros.
> 
> Use prb_first_seq() as the base for the 32bit seq macros instead
> because it is guaranteed to return 64bit sequence numbers without
> relying on any 32bit seq macros.
>
> Signed-off-by: John Ogness <john.ogness@linutronix.de>

This increases the risk of an overflow which would result
into wrong u64 sequence number. I was curious what numbers
we are talking about.

Here is my thinking:

The difference between prb_next_seq() and prb_first_seq() is
limited by the amount of messages stored in the buffer.

Every stored message must have meta data stored in desc_ring.
The current code expects average length of the message to be 32.

Now, __ulseq_to_u64seq() would give wrong result when
(s32)((u32)rb_first_seq - ulseq) overflows. It is 2^31
sequence numbers.

If we create desc buffer for these messages then the
data buffer would be 32x bigger, so:

   2^31 * 32 = 68,719,476,736 => 68GB

=> we seem to be on the safe side with a good reserve.


Reviewed-by: Petr Mladek <pmladek@suse.com>

Best Regards,
Petr