Re: [PATCH] x86: Fix earlyprintk ttyS2/3

[Ingo Molnar <mingo@kernel.org>]

* Daniel J Blueman <daniel@numascale.com> wrote:

> Quite a few platforms use ttyS2 for their serial-over-LAN, so fix early
> printk support for ttyS2 and 3, avoiding the need to hard-code the IO port.

Nice.

> 
> Signed-off-by: Daniel J Blueman <daniel@numascale.com>
> ---
>  arch/x86/boot/early_serial_console.c | 9 +++++----
>  1 file changed, 5 insertions(+), 4 deletions(-)
> 
> diff --git a/arch/x86/boot/early_serial_console.c b/arch/x86/boot/early_serial_console.c
> index 5df2869..4f4c2e6 100644
> --- a/arch/x86/boot/early_serial_console.c
> +++ b/arch/x86/boot/early_serial_console.c
> @@ -71,15 +71,16 @@ static void parse_earlyprintk(void)
>  			else
>  				pos = e - arg;
>  		} else if (!strncmp(arg + pos, "ttyS", 4)) {
> -			static const int bases[] = { 0x3f8, 0x2f8 };
> -			int idx = 0;
> +			static const int bases[] = {0x3f8, 0x2f8, 0x3e8, 0x2e8};
> +			unsigned idx = 0;
>  
>  			if (!strncmp(arg + pos, "ttyS", 4))
>  				pos += 4;

Btw., that's a weird pattern: why do the strncmp() twice? We already 
know that we matched, in this branch. Could be written as:

			pos += 4;

?

>  
> -			if (arg[pos++] == '1')
> -				idx = 1;
> +			if ((arg[pos] - '0') < (sizeof(bases) / sizeof(bases[0])))
> +				idx = arg[pos] - '0';

Nit: use ARRAY_SIZE()?

Not so nit: with your change, a typo in boot parameters like "ttyS-" 
would underflow the index and dereference randomly around bases[] with 
a negative index:

Also, higher indices like "ttyS4" will overflow beyond the end of the 
array.

Please add proper bounds checking and warning messages.

>  
> +			pos++;
>  			port = bases[idx];
>  		}

Thanks,

	Ingo