----- Original Message -----
----- Original Message -----
>
> > > > Given that those 3 members are all pointer types, you could just use
> > > > "sizeof(void *)" for the readmem() calls, as is done
throughout the
> > > > crash code. Note that there are not any MEMBER_SIZE_INIT() callers
> > > > for structure members that are known, guaranteed, pointer values.
> > >
> > > I'm doing this considering x86 build as x86_64 binary where pointer
size
> > > differs, or is this unnecessary for such x86 build? Sorry, I didn't
test
> > > x86 build as x86_64 binary.
> >
> > With "make target=x86", crash is built with -m32 to create a 32-bit
x86 binary
> > that runs on an x86_64 host, so kernel pointer sizes will always be the same
> > as the crash binary.
>
> I was confused this.
>
> I made the 2nd version of the patch and reflected your comments in it.
> Could you review this?
Hi Daisuke,
This one looks good. I'm going to tweak the help page output a bit, and only
initialize all of the sizes/offsets when the command is called the first
time.
I'll try to check it in tomorrow.
Thanks,
Dave
Hi Daisuke,
I made the changes referenced above, and added sk_buff_len to the size_table[] dump.
Queued for crash-7.1.9:
https://github.com/crash-utility/crash/commit/f4623a2f14b169bfd5e4d782bc4...
Thanks,
Dave