https://sourceware.org/bugzilla/show_bug.cgi?id=26556
Bug ID: 26556
Summary: relax_segment calculates incorrect target frag when
OCTETS_PER_BYTE > 1
Product: binutils
Version: 2.34
Status: UNCONFIRMED
Severity: minor
Priority: P2
Component: gas
Assignee: unassigned at sourceware dot org
Reporter: tuckkern+sourceware at gmail dot com
Target Milestone: ---
Created attachment 12810
--> https://sourceware.org/bugzilla/attachment.cgi?id=12810&action=edit
Possible patch
The case statement handling rs_org in write.c around line 2937 does not appear
to calculate the correct target fragment address when OCTETS_PER_BYTE > 1
It only applies OCTETS_PER_BYTE scaling to the symbol value, but not the
offset.
The end result is symbols in .bss with overlapping addresses.
e.g. from readelf - array_ has a size of 4, but var_ is located at address 2
Num: Value Size Type Bind Vis Ndx Name
5: 00000000 4 OBJECT LOCAL DEFAULT 4 array_
6: 00000002 1 OBJECT LOCAL DEFAULT 4 var_
The attached patch is a possible fix, but II am unsure if this is the correct
place to address the issue or if there is another root cause.
Using the above patch the symbol table now looks like so:
Num: Value Size Type Bind Vis Ndx Name
5: 00000000 4 OBJECT LOCAL DEFAULT 4 array_
6: 00000004 1 OBJECT LOCAL DEFAULT 4 var_
--
You are receiving this mail because:
You are on the CC list for the bug.
