Refactor - Separates the text section from read-only sections

[Lichtso]

This PR removes the indirection in which the text section is retrieved from the possibly concatenated read-only sections. There is only ever one text-section and no need to concatenate it. So instead the text section can be borrowed as a sub-slice of the relocated ELF bytes directly. Also removes the mem_size of text_section_info as that was double counted and the debug name, as it was unused.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 92.30769% with 1 line in your changes missing coverage. Please review.

Project coverage is 90.00%. Comparing base (a996494) to head (5749444).
Report is 1 commits behind head on main.

Files	Patch %	Lines
src/elf.rs	92.30%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #553      +/-   ##
==========================================
+ Coverage   89.92%   90.00%   +0.08%     
==========================================
  Files          22       22              
  Lines        9735     9685      -50     
==========================================
- Hits         8754     8717      -37     
+ Misses        981      968      -13     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[alessandrod]

There is only ever one text-section and no need to concatenate it

I haven't looked at the change yet but this is not true?

[Lichtso]

There is only ever one text-section and no need to concatenate it

We only consider the byte range of one section:

rbpf/src/elf.rs

Line 415 in 179a0f9

offset_range: text_section.file_range().unwrap_or_default(),

which is exactly named ".text":

rbpf/src/elf.rs

Line 396 in 179a0f9

let text_section = elf.section(b".text")?;

[LucasSte]

I'm guessing that you are adding to MM_PROGRAM_START because you subtracted it from the offset in the initialization to maintain parallelism between this function and the get_ro_region, right?

Would you mind explaining this in a comment?

[Lichtso]

rbpf/src/elf.rs

Line 237 in a400aa7

/// The first field is the offset of the section from MM_PROGRAM_START. The

The structs field is defined as being relative to MM_PROGRAM_START, both the text section and all read-only sections live in that MemoryRegion.

[LucasSte]

In line 469, you're subtracting the values

 text_section_vaddr.saturating_sub(ebpf::MM_PROGRAM_START) as usize,

My concern was that it was a little confusing that you subtract and then you add in get_text_bytes. A comment to explain why this is the case would be helpful.

[LucasSte]

Should we include an arm like this to account for a possible owned data (in the text section)?

[Lichtso]

The text section is not supposed to ever be owned / copied. Hence why all match expressions against it have an unreachable! arm.