Re: arm-none-eabi-as: what decides .text section alignment/padding?

["Richard Earnshaw (lists)" <Richard.Earnshaw@arm.com>]

On 19/06/2023 17:01, Alexandru Onea via Binutils wrote:
> Therefore, is it that the only difference between the two cases shown in my
> example boils down to whether the assembler sees an instruction or not?
> I understand it as follows:
> 
> Case A: .text section with no instruction within, the assembler says "this
> monkey asks me to put data in a .text section, whatever, maybe where he
> comes from .text means something else"
> Case B: .text section with data and at least one instruction, the assembler
> says "this monkey now crosses the line, if he wants instructions, I must
> make sure that at least other .text sections can be merged into it"
> 
> So does the assembler assume that if there are no instructions in a .text
> section, maybe it is not actually a .text section in the classical sense?

Alignment is taken from the most aligned element in the section.  It 
starts off at byte alignment and if you add an instruction with 
half-word alignment (thumb) it gets raised to that.  If you add an 
instruction with word alignment (arm) it gets raised again.  If you add 
an alignment directive with even higher alignment then that will 
dominate the overall alignment of the section.

The only difference between this and, say, the .data section is that the 
alignment is implied by the instruction rather than explicitly requiring 
an alignment directive.

A final note on this is that a .text section may contain both code and 
data.  But the two are not the same and the assembler needs to keep 
track of them (especially on big-endian systems) so that byte swapping 
of instructions can be done correctly at link time.  So don't try to 
fake up instructions by emitting .byte or similar directives, it will 
just lead to silent problems later on - use the .inst directive if you 
need to emit an opcode as binary.

R.

> 
> Best regards,
> Alexandru N. Onea
> 
> 
> On Mon, 19 Jun 2023 at 16:31, Richard Earnshaw (lists) <
> Richard.Earnshaw@arm.com> wrote:
> 
>> On 19/06/2023 08:14, Alexandru Onea via Binutils wrote:
>>> Hello, community!
>>>
>>> I am playing with arm-none-eabi-as trying to understand how it aligns
>>> sections. I have the following source:
>>>
>>> ; source.s
>>> .text
>>> .byte 0xff
>>> .byte 0xff
>>> .byte 0xff
>>>
>>> I am inspecting the resulting object file:
>>>
>>> $ arm-none-eabi-as -mthumb -o source.o source.s
>>> $ arm-none-eabi-readelf -S source.o
>>> There are 8 section headers, starting at offset 0xec:
>>>
>>> Section Headers:
>>>     [Nr] Name              Type            Addr     Off    Size   ES Flg
>> Lk Inf Al
>>>     [ 0]                   NULL            00000000 000000 000000 00
>>>    0   0  0*  [ 1] .text             PROGBITS        00000000 000034
>>> 000003 00  AX  0   0  1
>>> *  [ 2] .data             PROGBITS        00000000 000037 000000 00
>>> WA  0   0  1
>>>     [ 3] .bss              NOBITS          00000000 000037 000000 00  WA
>> 0   0  1
>>>     [ 4] .ARM.attributes   ARM_ATTRIBUTES  00000000 000037 000014 00
>> 0   0  1
>>>     [ 5] .symtab           SYMTAB          00000000 00004c 000060 10
>> 6   6  4
>>>     [ 6] .strtab           STRTAB          00000000 0000ac 000004 00
>> 0   0  1
>>>     [ 7] .shstrtab         STRTAB          00000000 0000b0 00003c 00
>> 0   0  1
>>>
>>> The .text section is byte-aligned and contains the 3 bytes.
>>>
>>> Now, I add an instruction to source.s:
>>>
>>> ; source.s
>>> .text
>>> .byte 0xff
>>> nop
>>> .byte 0xff
>>> .byte 0xff
>>>
>>> Looking into the object file, now all of a sudden the .text section is
>>> halfword-aligned:
>>>
>>> There are 8 section headers, starting at offset 0x114:
>>>
>>> Section Headers:
>>>     [Nr] Name              Type            Addr     Off    Size   ES Flg
>> Lk Inf Al
>>>     [ 0]                   NULL            00000000 000000 000000 00
>>>    0   0  0*  [ 1] .text             PROGBITS        00000000 000034
>>> 000006 00  AX  0   0  2
>>> *  [ 2] .data             PROGBITS        00000000 00003a 000000 00
>>> WA  0   0  1
>>>     [ 3] .bss              NOBITS          00000000 00003a 000000 00  WA
>> 0   0  1
>>>     [ 4] .ARM.attributes   ARM_ATTRIBUTES  00000000 00003a 000014 00
>> 0   0  1
>>>     [ 5] .symtab           SYMTAB          00000000 000050 000080 10
>> 6   8  4
>>>     [ 6] .strtab           STRTAB          00000000 0000d0 000007 00
>> 0   0  1
>>>     [ 7] .shstrtab         STRTAB          00000000 0000d7 00003c 00
>> 0   0  1
>>>
>>> What is causing the assembler to decide to pad the section in the second
>>> case? I am confused because:
>>>
>>>      1. if the section is .data then the assembler will not pad it anyway,
>>>      which makes sense, but
>>>      2. even if the section is .text, the assembler won't pad it unless it
>>>      sees an instruction (I can have as many data directives, the section
>> won't
>>>      be padded without having also an instruction), and finally
>>>      3. the nop instruction is definitely not aligned and the assembler
>> has
>>>      no problem with it, but it still decides to care about section
>> alignment.
>>>
>>> How is the assembler deciding here to align and pad the section? Can I
>>> force the assembler to not pad the .text section even if I have an
>>> instruction?
>>
>> You've told the assembler to interpret the contents of the source file
>> as containing thumb instructions.  Thumb instructions must be (at least)
>> two byte aligned (the fact that your source file deliberately misaligns
>> the instruction is a bug in your code).  If the section were not aligned
>> then the linker would not be able to correctly merge consecutive
>> sections if you had a mix of data-only text sections (like your first
>> example) and sections containing code.
>>
>> Note that if you had used -marm instead of -mthumb the alignment would
>> be set to 4 as all 'arm' instructions must be 4-byte aligned.
>>
>> R.
>>
>>