1484406 - rust-size doesn't understand lld-link binaries

Status: RESOLVED FIXED

(Firefox Build System :: General, defect)

Description

[(Away)]

It looks like something is coming back as 0xffffffff here: https://treeherder.mozilla.org/perf.html#/graphs?timerange=7776000&series=try,1686353,1,2

Comment 1

[(Away)]

Or maybe this subtraction does something surprising https://github.com/luser/rust-size/blob/master/src/main.rs#L97

Comment 2

[Eric Rahm [:erahm]]

Example output for xul.dll:

> 04:12:22     INFO - Getting output from command: ['z:\\build\\build\\src\\rust-size\\rust-size.exe', 'z:\\build\\build\\src\\obj-firefox\\dist\\bin\\xul.dll']
> 04:12:22     INFO - Copy/paste: z:\build\build\src\rust-size\rust-size.exe z:\build\build\src\obj-firefox\dist\bin\xul.dll
> 04:12:22     INFO - Reading from file tmpfile_stdout
> 04:12:22     INFO - Output received:
> 04:12:22     INFO -  {
> 04:12:22     INFO -    "Text": {
> 04:12:22     INFO -      ".00cfg": 32,
> 04:12:22     INFO -      ".gfids": 56,
> 04:12:22     INFO -      ".orpc": 46,
> 04:12:22     INFO -      ".pdata": 1578348,
> 04:12:22     INFO -      ".rdata": 10756468,
> 04:12:22     INFO -      ".reloc": 651380,
> 04:12:22     INFO -      ".rodata": 11708384,
> 04:12:22     INFO -      ".rsrc": 9424,
> 04:12:22     INFO -      ".text": 67240293,
> 04:12:22     INFO -      "_text32": 220440
> 04:12:22     INFO -    },
> 04:12:22     INFO -    "Data": {
> 04:12:22     INFO -      ".data": 270336,
> 04:12:22     INFO -      ".tls": 1024
> 04:12:22     INFO -    },
> 04:12:22     INFO -    "Bss": {
> 04:12:22     INFO -      ".bss": 4295580490
> 04:12:22     INFO -    }
> 04:12:22     INFO -  }

Seems okay for other binaries:

> 04:12:22     INFO - Getting output from command: ['z:\\build\\build\\src\\rust-size\\rust-size.exe', 'z:\\build\\build\\src\\obj-firefox\\dist\\bin\\mozavutil.dll']
> 04:12:22     INFO - Copy/paste: z:\build\build\src\rust-size\rust-size.exe z:\build\build\src\obj-firefox\dist\bin\mozavutil.dll
> 04:12:22     INFO - Reading from file tmpfile_stdout
> 04:12:22     INFO - Output received:
> 04:12:22     INFO -  {
> 04:12:22     INFO -    "Text": {
> 04:12:22     INFO -      ".00cfg": 32,
> 04:12:22     INFO -      ".gfids": 8,
> 04:12:22     INFO -      ".pdata": 3564,
> 04:12:22     INFO -      ".rdata": 98656,
> 04:12:22     INFO -      ".reloc": 1824,
> 04:12:22     INFO -      ".rodata": 32,
> 04:12:22     INFO -      ".rsrc": 896,
> 04:12:22     INFO -      ".text": 107638
> 04:12:22     INFO -    },
> 04:12:22     INFO -    "Data": {
> 04:12:22     INFO -      ".data": 512
> 04:12:22     INFO -    },
> 04:12:22     INFO -    "Bss": {
> 04:12:22     INFO -      ".bss": 55468
> 04:12:22     INFO -    }
> 04:12:22     INFO -  }

Comment 3

[Eric Rahm [:erahm]]

Gerald ran into an issue locally which helps narrow things down. It looks like an issue with subtraction as David mentioned in comment 1. It's surprising we have a smaller virtual size for a writable chunk. I'll see if I can narrow down which section it is locally.

> $ RUST_BACKTRACE=1 cargo run ..\\..\\obj-mc-dbg\\dist\\bin\\xul.dll
>     Finished dev [unoptimized + debuginfo] target(s) in 0.10s
>      Running `target\debug\rust-size.exe '..\..\obj-mc-dbg\dist\bin\xul.dll'`
> thread 'main' panicked at 'attempt to subtract with overflow', src\main.rs:97:33
> stack backtrace:
>    [...]
>    9: core::panicking::panic
>              at C:\projects\rust\src\libcore\panicking.rs:53
>   10: rust_size::sections::{{closure}}
>              at .\src\main.rs:97
>   11: core::ops::function::impls::{{impl}}::call_once<(goblin::pe::section_table::SectionTable*),closure>
>              at C:\projects\rust\src\libcore\ops\function.rs:271
>   12: core::option::Option<goblin::pe::section_table::SectionTable*>::map<goblin::pe::section_table::SectionTable*,(alloc::string::String, u64, rust_size::Section),mut closure*>
>              at C:\projects\rust\src\libcore\option.rs:414
>   13: core::iter::{{impl}}::next<(alloc::string::String, u64, rust_size::Section),core::slice::Iter<goblin::pe::section_table::SectionTable>,closure>
>              at C:\projects\rust\src\libcore\iter\mod.rs:1326
>   14: alloc::vec::{{impl}}::spec_extend<(alloc::string::String, u64, rust_size::Section),core::iter::Map<core::slice::Iter<goblin::pe::section_table::SectionTable>, closure>>
>              at C:\projects\rust\src\liballoc\vec.rs:1880
>   15: alloc::vec::{{impl}}::from_iter<(alloc::string::String, u64, rust_size::Section),core::iter::Map<core::slice::Iter<goblin::pe::section_table::SectionTable>, closure>>
>              at C:\projects\rust\src\liballoc\vec.rs:1863
>   16: alloc::vec::{{impl}}::from_iter<(alloc::string::String, u64, rust_size::Section),core::iter::Map<core::slice::Iter<goblin::pe::section_table::SectionTable>, closure>>
>              at C:\projects\rust\src\liballoc\vec.rs:1749
>   17: core::iter::iterator::Iterator::collect<core::iter::Map<core::slice::Iter<goblin::pe::section_table::SectionTable>, closure>,alloc::vec::Vec<(alloc::string::String, u64, rust_size::Section)>>
>              at C:\projects\rust\src\libcore\iter\iterator.rs:1419
>   18: rust_size::sections
>              at .\src\main.rs:89
>   19: rust_size::real_main
>              at .\src\main.rs:160
>   20: rust_size::main
>              at .\src\main.rs:170
> [...]
> error: process didn't exit successfully: `target\debug\rust-size.exe ..\..\obj-mc-dbg\dist\bin\xul.dll` (exit code: 101)

Comment 4

[Eric Rahm [:erahm]]

It appears to be choking on the '.tls' section which has an on disk size of 1024 and in memory size of 986. Running `gsize` on it just shows a value of 0 for .bss.

Comment 5

[Eric Rahm [:erahm]]

Attachment: Update rust-size version

This updates the rust-size build to reference the latest version [1] that
includes a work-around for calculating the .bss of a PE file that has a
writable section with a smaller on-disk size than in-memory size.

[1] https://github.com/luser/rust-size/commit/ab659d93b1faba95307df952aefe3fbed3583669

Comment 6

[Mike Hommey [:glandium]]

Comment on attachment 9007095 [details] [diff] [review]
Update rust-size version

Review of attachment 9007095 [details] [diff] [review]:
-----------------------------------------------------------------

Relatedly, could we move this repo under the mozilla project on github?

Comment 7

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/10ffdec164a9
Update rust-size version. r=glandium

Comment 8

[Eric Rahm [:erahm]]

(In reply to Mike Hommey [:glandium] from comment #6)
> Comment on attachment 9007095 [details] [diff] [review]
> Update rust-size version
> 
> Review of attachment 9007095 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> Relatedly, could we move this repo under the mozilla project on github?

Filed https://github.com/luser/rust-size/issues/2, presumably Ted can move it.

Comment 9

[Daniel Varga [:dvarga]]

https://hg.mozilla.org/mozilla-central/rev/10ffdec164a9