This is the mail archive of the binutils@sources.redhat.com mailing list for the binutils project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: PATCH: BZ/175: Fix dynamic string offset


On Fri, 2004-05-28 at 22:40, Alan Modra wrote:
> Well, the particular part that I wanted you to comment on related to
> ia64 relocations. 

I can tell you lots of stuff you don't want to know about IA-64
assembler features (and mis-features).  But I have limited knowledge of
IA-64 relocations and limited knowledge of how bfd works.

HJ's example uses an LTOFF22X relocation, there is nothing particularly
special about that.  I don't think you can determine anything special by
looking at the relocation here.

HJ's example looks like it is generating an address outside an object,
which is only valid if it is later given an offset that puts us back
inside the object again.  In this case either we shouldn't be generating
an error for out-of-bounds addresses, or else we shouldn't be generating
out-of-bounds addresses for mergeable sections.  The second one implies
a compiler fix, probably not merging offsets with symbols in a CONST if
the offset takes us outside the object, and the object is in a mergable
section.  The first one implies that we need to differentiate between
internal offsets and programmer supplied offsets.  Internal offsets give
an error if they go outside the mergable section because that indicates
a bug, but programmer supplied offsets do not, because the programmer
meant to do that.  I think one of HJ's patches tried to implement
something like this.
-- 
Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]