[Bug ld/27566] [RISC-V] relocation truncated to fit: R_RISCV_GPREL_I against aymbol

[wilson at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27566

Jim Wilson <wilson at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |wilson at gcc dot gnu.org

--- Comment #5 from Jim Wilson <wilson at gcc dot gnu.org> ---
The c.lui case was fixed because a bug turned up in regression testing, and
Ilya Diachkov debugged it and wrote a patch.  The tail end of the thread is
here
https://sourceware.org/pipermail/binutils/2019-July/107702.html
Looks like I missed the fact that other cases needed a similar fix.  And I
didn't notice the MAXPAGESIZE versus COMMONPAGESIZE issue.  it does look like
fixing all 3 places to use MAXPAGESIZE+COMMONPAGESIZE is the correct fix.

GP is computed so that __DATA_BEGIN__+0x800 is the minimum value, so that it
can never overlap rodata section, but apparently that isn't enough in this
case.  Maybe gp or __DATA__BEGIN__ aren't being updated until after we use them
because of how relaxation works.

Yes, we can add symbols to the linker script if necessary, but __DATA_BEGIN__
is alraedy in the right place and we should be able to use that.  Except it
looks like it might not work in this case.

-- 
You are receiving this mail because:
You are on the CC list for the bug.