-
Notifications
You must be signed in to change notification settings - Fork 161
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
elf: fix is_import to check st_shndx #436
base: master
Are you sure you want to change the base?
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
As mentioned in comment, swapping out these functions to suddenly use shndx doesn't seem right to me; adding the STB_WEAK check in the is_import function is good catch (though it now has me wondering whether all these functions should be marked private).
However, we can review adding a new api that uses the shndx in another patch if that is something people really want, or some other design that incorporates it somehow.
@@ -94,9 +94,9 @@ pub fn st_visibility(other: u8) -> u8 { | |||
|
|||
/// Is this information defining an import? | |||
#[inline] | |||
pub fn is_import(info: u8, value: u64) -> bool { | |||
pub fn is_import(info: u8, shndx: u64) -> bool { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
idly wondering if we should de-pub this along with the other functions; i wonder why they were made pub
, could have been a historical accident or perhaps someone uses them for whatever reason; i don't see bingrep for example using it anywhere
let bind = st_bind(info); | ||
bind == STB_GLOBAL && value == 0 | ||
(bind == STB_GLOBAL || bind == STB_WEAK) && shndx == SHN_UNDEF as _ |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
this is a good catch (the missing STB_WEAK), but changing this function to operate semantically on shndx
values instead of st_value
, while backwards compatible, has wider implications.
for example, any clients relying on st_value
being used in these computations will now break when section headers are stripped.
imho my general thesis that what a dynamic linker can import is how we've defined import, while arguably more nuanced than this, is a good one and has value (no pun intended). using st_value
is a heuristic for that thesis.
that being said, it seems like people are hitting some issues with this, so we can consider adding something like is_import_from_sections
or something which is new api to opt into using the section header (perhaps along with st_value) as the determinant.
Closes #288
Closes #419
I also found this bug on with ELF produced on illumos.
st_value
may not be zero because it may indicate the offset of a import function.