"obf name not found, this should never happen" when build with obfuscation enabled

[bencinn]

Describe the bug

obf name "main.xx" not found, this should never happen when building with -obf flags

Reproduction Steps

test.v:

fn add(a int, b int) int {
	return a + b
}

fn main() {
	println(add(1, 3))
}

v -o test -autofree -show-timings -obf -skip-unused -prod -arch x64 -os linux -cc gcc test.v
Edit: minimized the reproduction build step.
v -o test -obf test.v

Expected Behavior

It build

Current Behavior

54.470   ms Builder.front_stages.parse_files
0.083    ms Builder.resolve_deps
21.476   ms Builder.parse_imports
19.259   ms SCAN
52.204   ms PARSE
75.996   ms ALL_FRONT_STAGES
0.003    ms Checker.generic_insts_to_concrete
23.720   ms CHECK
4.851    ms TRANSFORM
0.037    ms Table.complete_interface_check
1.404    ms all_fn_const_and_global
2.410    ms mark_used
0.576    ms Gen.sort_structs
0.957    ms cgen init
V panic: cgen: fn_decl: obf name "main.add" not found, this should never happen
v hash: 99e0832

Possible Solution

No response

Additional Information/Context

No response

V version

0.4.2 00165d1

Environment details (OS name and version, etc.)

V full version: V 0.4.2 714a64d.00165d1
OS: linux, "Gentoo Linux"
Processor: 4 cpus, 64bit, little endian, Intel(R) Core(TM) i5-7200U CPU @ 2.50GHz

getwd: /home/rori/code/stuff
vexe: /home/rori/code/v/v
vexe mtime: 2023-10-29 09:53:09

vroot: OK, value: /home/rori/code/v
VMODULES: OK, value: /home/rori/.vmodules
VTMP: OK, value: /tmp/v_1000

Git version: git version 2.41.0
Git vroot status: weekly.2023.43-43-g00165d14
.git/config present: true

CC version: cc (Gentoo 13.2.1_p20230826 p7) 13.2.1 20230826
thirdparty/tcc status: thirdparty-linux-amd64 12f392c3

Note

You can use the 👍 reaction to increase the issue's priority for developers.

Please note that only the 👍 reaction to the issue itself counts as a vote.
Other reactions and those to comments will not be taken into account.

[bencinn]

After I tried debugging, it seems that the g.obf_table is always empty?
I tried this.

// vlib/v/gen/c/cgen.v
	if g.pref.obfuscate {
		mut i := 0
		// fns
		for key, f in g.table.fns {
			println('key: ${key}, f: ${f}')
			if f.mod != 'main' &&  key != 'main' {
				continue
			}
			println('inserting key ${key} into the obf_table?')
			g.obf_table[key] = '_f${i}'
			i++
		}
		// ...

// vlib/v/gen/c/fn.v
if g.pref.obfuscate && g.cur_mod.name == 'main' && name.starts_with('main__') && !node.is_main
		&& node.name != 'str' {
		mut key := node.name
		if node.is_method {
			sym := g.table.sym(node.receiver.typ)
			key = sym.name + '.' + node.name
		}
		g.writeln('/* obf: ${key} */')
		name = g.obf_table[key] or {
	println('obf_table: ${g.obf_table}')
				println('${node}')
			panic('cgen: fn_decl: obf name "${key}" not found, this should never happen')
		}
	}
// ...

inserting key main.add into the obf_table?
...
obf_table: {}

Race condition?

[JalonSolov]

All you need to show the error is v -obf test.v (since -o test is the default... it uses the base name of the file as the output filename).

[bencinn]

All you need to show the error is v -obf test.v (since -o test is the default... it uses the base name of the file as the output filename).

Thanks!

[bencinn]

Ok more debug information.
It seems that the map obf_table got eaten/overwritten/deleted?

// vlib/v/gen/c/cgen.v
for key, f in g.table.fns {
			println('key: ${key}, f: ${f}')
			if f.mod != 'main' && key != 'main' {
				continue
			}
			println('inserting key ${key} into the obf_table?')
			g.obf_table[key] = '_f${i}'
			println('table after inserting key ${key}: ${g.obf_table}')
			i++
		}
// ...

table after inserting key main.main: {'main.add': '_f0', 'main.main': '_f1'}
...
obf_table: {}
...

[medvednikov]

Good find, looks we are not testing -obf, need to add it to CI.