Skip to content

Elf loading bug? #9

@cmcl

Description

@cmcl

I've been studying the elf loading procedure for SOS processes defined in elf.c. I think I've stumbled upon a bug: if the mapping fails, the frame cap is deleted and the slot is freed (the frame is also freed but is still mapped in SOS) however the code continues to use these afterwards which I expect to result in an exception around here where an attempt is made to use the frame cap.

The comment is a bit mysterious -- when it mentions all frames being mapped in, I assume it is talking about the SOS address space, but why "continue on to do the write" on a frame that's not mapped into the process' address space?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions