[edk2-devel] Linker scripts use of "-z common-page-size=0x20" etc.

[Ard Biesheuvel]

On Mon, 3 Apr 2023 at 22:33, Rebecca Cran <rebecca at bsdio.com> wrote:
>
> As part of my work on the toolchain definitions, I've come across a
> situation where ld.lld fails to align sections correctly, due to it
> being invoked via clang with the '-n' option, which causes GenFw to fail
> with "Section address not aligned to its own alignment.".
>

Stupid question: if it breaks stuff, why do you use -n ?

> The following messages are printed:
>
>
> ld.lld: warning: -z common-page-size set, but paging disabled by omagic
> or nmagic ld.lld: warning: address (0x558) of section .data is not a
> multiple of alignment (16)
>
> I tracked the problem down to GccBase.lds and ClangBase.lds, which have:
>
> /* * The alignment of the .data section should be less than or equal to
> the * alignment of the .text section. This ensures that the relative
> offset * between these sections is the same in the ELF and the PE/COFF
> versions of * this binary. */ .data ALIGN(ALIGNOF(.text)) :
> ALIGN(CONSTANT(COMMONPAGESIZE)) { *(.data .data.* .gnu.linkonce.d.*)
> *(.bss .bss.*) }
>
> I can work around the problem by removing "ALIGN(ALIGNOF(.text)", but
> I'm wondering if our use of COMMONPAGESIZE/MAXPAGESIZE is correct.
>

What do you mean by 'correct'? The intent is clearly to declare the
mapping granule size, and for SEC/PEI binaries that execute in place
from flash, the MMU page size is not the most useful quantum here.

>
> We pass in a value of 0x20, 0x40 or 0x1000 to "-z common-page-size" with
> 0x20 being the most common value.
>
> Given the page size of the target will never be 32 bytes, the following
> comment on https://reviews.llvm.org/D61688 makes sense:
>
>
> "There is at least one linkerscript in Tianocore edk2 that (ab)uses -n
> -z common-page-size=0x20 to use CONSTANT(COMMONPAGESIZE) as if it were a
> preprocessor macro set with -D in the compiler. The usual approach to
> this is to pre-process the linkerscript."
>

'Page size' is highly context specific, and toolchain people are
(imho) usually quite quick to call something abuse if it does not
match their narrow definition of how a compiler or linker should be
used. For the same reason, it has been so difficult to get a compiler
to understand that the desire for position independent code does not
imply that we want GOTs, or care about ELF symbol preemption or
copy-on-write footprint of relocatable segments. In general, the bare
metal use case (which includes EFI) is quite poorly understood by many
people working on toolchains.

>
> I'm wondering what the correct approach is here: should we do something
> similar to how we set PECOFF_HEADER_SIZE and define a SECTION_ALIGNMENT
> symbol?

We cannot, that is the reason for using the page size switches here:
using a symbol to set the location pointer is fine, but using a symbol
to set the alignment of a section is not.

> Or, as discussed on Discord should we just use
> CONSTANT(MAXPAGESIZE) and ignore how it's normally used to specify the
> maximum allowable page size for an executable?
>

Note that (when I last checked), the only effect of setting -z
xxx-page-size is that those macros assume the associated value in the
linker script. Nothing else in the linker changes behavior (with the
exception of the warning you are seeing)

So claiming abuse because the provided value does not match the page
size of an OS that might also run on the same system is strenuous, and
I think our use of it is fine. AIUI, the reason for having
ClangBase.lds in addition to GccBase.lds is the fact that LLD does not
support common but only max page size, so I think it should be fine to
merge the two, and use max-page-size everywhere.


-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#102463): https://edk2.groups.io/g/devel/message/102463
Mute This Topic: https://groups.io/mt/98045342/1813853
Group Owner: devel+owner at edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [edk2-devel-archive at redhat.com]
-=-=-=-=-=-=-=-=-=-=-=-