[Cbe-oss-dev] [PATCH 1/1] Fix another user-visible SPU coredump bug

[Arnd Bergmann]

On Tuesday 21 August 2007, Michael Ellerman wrote:
> > This means that the notes sizes were correct, but the content
> > was incorrectly written in binary form instead of using the
> > correct format string.
> 
> No they weren't. They might have been correct for certain values, but
> there's no guarantee what size the output of:
> 
>  sprintf(s, "0x%lx", unknown_value)
> 
> will be.
> 
> To support that we're going to need to calculate the size on the fly,
> based on the content of the file, rather than hard coding. Which means
> completely redesigning the reading logic basically.

I thought that the size values in there were the maximum sizes that
we can fit in. AFAICT, we can always pad the space with zeroes without
causing problems for programs that interpret the dump, so there should
be no need to calculate the size on the fly.

	Arnd <><