On Thu Jun 15, 2023 at 1:47 AM CDT, Jan Beulich wrote:
> On 14.06.2023 18:36, Shawn Anastasio wrote:
> > On Wed Jun 14, 2023 at 10:51 AM CDT, Jan Beulich wrote:
> >> On 13.06.2023 16:50, Shawn Anastasio wrote:
> >>> + DECL_SECTION(.bss) { /* BSS */
> >>> + __bss_start = .;
> >>> + *(.bss.stack_aligned)
> >>> + . = ALIGN(PAGE_SIZE);
> >>> + *(.bss.page_aligned)
> >>
> >> ... the one between the two .bss parts looks unmotivated. Within
> >> a section definition ALIGN() typically only makes sense when followed
> >> by a label definition, like ...
> >
> > Correct me if I'm wrong, but wouldn't the ALIGN here serve to ensure
> > that the subsequent '.bss.page_aligned' section has the correct alignment
> > that its name implies?
>
> Yes and no. Thing is that every contribution to .bss.page_aligned already
> needs to specify page alignment itself, or else it may break if any earlier
> contribution was page-aligned, but not a full page in size (which I think
> the compiler wouldn't allow to happen, but assembly code can result in
> such). Note how this very ALIGN() was recently dropped from RISC-V code,
> and my respective Arm side patch is also about to go in.
That makes sense, thanks. I'll get rid of the ALIGN here in v4.
> Jan
Thanks,
Shawn
