Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

runtime: efence=1 and cgocheck=2 are not mutually compatible #42724

bradfitz opened this issue Nov 19, 2020 · 1 comment

runtime: efence=1 and cgocheck=2 are not mutually compatible #42724

bradfitz opened this issue Nov 19, 2020 · 1 comment


Copy link

@bradfitz bradfitz commented Nov 19, 2020

Low priority, but I discovered that GODEBUG=efence=1,cgocheck=2 together are not compatible.

A simple hello world using os/exec fails like:

write of Go pointer 0xc000113c00 to non-Go memory 0x2b36ee9f9a0
fatal error: Go pointer stored into non-Go memory

runtime stack:
runtime.throw(0x221c74, 0x24)
        /home/bradfitz/sdk/go1.15/src/runtime/panic.go:1116 +0x79
        /home/bradfitz/sdk/go1.15/src/runtime/cgocheck.go:55 +0xb0
        /home/bradfitz/sdk/go1.15/src/runtime/asm_amd64.s:370 +0x6b

goroutine 1 [running]:
        /home/bradfitz/sdk/go1.15/src/runtime/asm_amd64.s:330 fp=0x2b36ee9f7b8 sp=0x2b36ee9f7b0 pc=0xc7220
runtime.cgoCheckWriteBarrier(0x2b36ee9f9a0, 0xc000113c00)
        /home/bradfitz/sdk/go1.15/src/runtime/cgocheck.go:53 +0xd6 fp=0x2b36ee9f7f0 sp=0x2b36ee9f7b8 pc=0x64c36
runtime.wbBufFlush(0x2b36ee9f9a0, 0xc000113c00)
        /home/bradfitz/sdk/go1.15/src/runtime/mwbbuf.go:196 +0x8a fp=0x2b36ee9f820 sp=0x2b36ee9f7f0 pc=0x90cca
runtime.gcWriteBarrier(0x6f115, 0x1e92e0, 0x2b36ee9f803, 0x0, 0xc000113c00, 0x0, 0x351501, 0xc000168580, 0x2b36ee9fa90, 0x1be5ee, ...)
        /home/bradfitz/sdk/go1.15/src/runtime/asm_amd64.s:1461 +0xb5 fp=0x2b36ee9f8a8 sp=0x2b36ee9f820 pc=0xc90b5
runtime.gcWriteBarrierCX(0x1e92e0, 0x2b36ee9f803, 0x0, 0xc000113c00, 0x0, 0x351501, 0xc000168580, 0x2b36ee9fa90, 0x1be5ee, 0x1e92e0, ...)
        /home/bradfitz/sdk/go1.15/src/runtime/asm_amd64.s:1481 +0x7 fp=0x2b36ee9f8b0 sp=0x2b36ee9f8a8 pc=0xc9107
runtime.makemap(0x1e92e0, 0x2c, 0x2b36ee9f990, 0xc000168580)
        /home/bradfitz/sdk/go1.15/src/runtime/map.go:328 +0x195 fp=0x2b36ee9f8f8 sp=0x2b36ee9f8b0 pc=0x6f115
os/exec.dedupEnvCase(0x1, 0xc000110e00, 0x2c, 0x32, 0x0, 0x0, 0x0)
        /home/bradfitz/sdk/go1.15/src/os/exec/exec.go:749 +0xee fp=0x2b36ee9faa0 sp=0x2b36ee9f8f8 pc=0x1be5ee
os/exec.(*Cmd).Start(0xc000166580, 0x0, 0x0)
        /home/bradfitz/sdk/go1.15/src/os/exec/exec.go:425 +0x499 fp=0x2b36ee9fbe8 sp=0x2b36ee9faa0 pc=0x1bd839
os/exec.(*Cmd).Run(0xc000166580, 0xc00017ac30, 0xc000166580)
        /home/bradfitz/sdk/go1.15/src/os/exec/exec.go:338 +0x32 fp=0x2b36ee9fc10 sp=0x2b36ee9fbe8 pc=0x1bced2
os/exec.(*Cmd).CombinedOutput(0xc000166580, 0x3, 0x2b36ee9fd30, 0x1, 0x1, 0xc000166580)
        /home/bradfitz/sdk/go1.15/src/os/exec/exec.go:567 +0x9b fp=0x2b36ee9fc40 sp=0x2b36ee9fc10 pc=0x1be3bb
        /home/bradfitz/src/ +0xad fp=0x2b36ee9fd90 sp=0x2b36ee9fc40 pc=0x1c9b8d
        /home/bradfitz/src/ +0xd0 fp=0x2b36ee9ff88 sp=0x2b36ee9fd90 pc=0x1ca370
        /home/bradfitz/sdk/go1.15/src/runtime/proc.go:204 +0x209 fp=0x2b36ee9ffe0 sp=0x2b36ee9ff88 pc=0x9a149
        /home/bradfitz/sdk/go1.15/src/runtime/asm_amd64.s:1374 +0x1 fp=0x2b36ee9ffe8 sp=0x2b36ee9ffe0 pc=0xc8fe1

@ianlancetaylor said elsewhere:

Looks like with efence=1 stacks are allocated using sysAlloc. That
does not look like a Go pointer to the cgocheck code. The effect is
that when writing a Go pointer to the stack, it looks like writing a
Go pointer to non-Go memory, which cgocheck crashes on.

I don't see a simple fix, other than effectively disabling cgocheck
when using efence. You could open an issue, I guess, but I agree that
it doesn't seem like a big concern.

Another fix is just having the runtime throw early with an error message if those options are used together.

@cagedmantis cagedmantis added this to the Backlog milestone Nov 30, 2020
Copy link

@cagedmantis cagedmantis commented Nov 30, 2020

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
None yet
Linked pull requests

Successfully merging a pull request may close this issue.

None yet
2 participants