https://sourceware.org/bugzilla/show_bug.cgi?id=32460
--- Comment #4 from Stas Sergeev ---
Created attachment 15850
--> https://sourceware.org/bugzilla/attachment.cgi?id=15850&action=edit
my ld.bfd
(In reply to H.J. Lu from comment #3)
> Which linker were you using? I tried, ld, lld, gold and
https://sourceware.org/bugzilla/show_bug.cgi?id=32460
--- Comment #7 from Stas Sergeev ---
Created attachment 15851
--> https://sourceware.org/bugzilla/attachment.cgi?id=15851&action=edit
my libs
It is a fedora linker indeed.
--
You are receiving this mail because:
You are on the CC list for
https://sourceware.org/bugzilla/show_bug.cgi?id=32460
--- Comment #11 from Stas Sergeev ---
(In reply to H.J. Lu from comment #3)
> Which linker were you using? I tried, ld, lld, gold and mold. All worked.
Complete off-topic: this is the first
time I've heard of mold. Having the new
linker is e
https://sourceware.org/bugzilla/show_bug.cgi?id=32460
--- Comment #12 from Stas Sergeev ---
Ok, mold simply doesn't support
linker scripts. Whether or not it
supports STT_RELC is probably then
irrelevant.
--
You are receiving this mail because:
You are on the CC list for the bug.
https://sourceware.org/bugzilla/show_bug.cgi?id=32463
--- Comment #2 from Stas Sergeev ---
Hi Nick.
I probably forgot to mention the
most important thing. If you change:
TTT = .;
_LGROUP = TTT;
into:
_LGROUP = .;
then suddenly bfg linker gives:
R _LGROUP
... same as lld.
So the bug
https://sourceware.org/bugzilla/show_bug.cgi?id=32463
--- Comment #4 from Stas Sergeev ---
(In reply to Nick Clifton from comment #3)
> I think that this is probably a case where it would be very unwise to change
> the behaviour of the linker - since there are bound to be scripts that
> depend u
https://sourceware.org/bugzilla/show_bug.cgi?id=32463
--- Comment #5 from Stas Sergeev ---
(In reply to Nick Clifton from comment #1)
> If you change the kernel.ld script to look like this:
>
> SECTIONS
> {
> TTT = .;
> #SHOULD_FAIL = . + .;
> . = SIZEOF_HEADERS;
> . = ALIGN(
https://sourceware.org/bugzilla/show_bug.cgi?id=32463
--- Comment #7 from Stas Sergeev ---
(In reply to Nick Clifton from comment #6)
> Oh well, I will look into it next year ... :-}
Omg. :)
Happy New year!
This problem is not fatal for me,
it just popped up as a result of
a routine testing (as
https://sourceware.org/bugzilla/show_bug.cgi?id=32787
--- Comment #4 from Stas Sergeev ---
Thanks for the quick fix!
--
You are receiving this mail because:
You are on the CC list for the bug.
Assignee: unassigned at sourceware dot org
Reporter: stsp at users dot sourceforge.net
Target Milestone: ---
Created attachment 15993
--> https://sourceware.org/bugzilla/attachment.cgi?id=15993&action=edit
test case
Attached is a test case.
Please run test.sh:
$ ./
https://sourceware.org/bugzilla/show_bug.cgi?id=32950
--- Comment #10 from Stas Sergeev ---
Hmm, seems it has --libs-only-l and --libs-only-other,
so that works too.
Thanks!
--
You are receiving this mail because:
You are on the CC list for the bug.
https://sourceware.org/bugzilla/show_bug.cgi?id=32950
--- Comment #9 from Stas Sergeev ---
To clarify: I used to do
LDFLAGS=`pkg-config --libs`
because I though LDFLAGS supports
both things. `pkg-config --libs`
gives lots of ldflags, plus -lXX stuff.
If we are to split this, it would
be good if p
Assignee: unassigned at sourceware dot org
Reporter: stsp at users dot sourceforge.net
Target Milestone: ---
Created attachment 16077
--> https://sourceware.org/bugzilla/attachment.cgi?id=16077&action=edit
test case
Here is the fully automated
test-case, just run `make`:
c
https://sourceware.org/bugzilla/show_bug.cgi?id=32950
--- Comment #2 from Stas Sergeev ---
But I can't even change the link
order: configure puts LDFLAGS before
conftest.c, so all configure tests
start to fail when --as-needed is
added to LDFLAGS.
What would you suggest as a fix then?
--
You a
https://sourceware.org/bugzilla/show_bug.cgi?id=32950
--- Comment #3 from Stas Sergeev ---
Maybe gcc, when compiling directly
from .c file, should put the intermediate
object at the beginning of the linker cmdline?
--
You are receiving this mail because:
You are on the CC list for the bug.
https://sourceware.org/bugzilla/show_bug.cgi?id=32950
--- Comment #6 from Stas Sergeev ---
So in my case I am providing (via LDFLAGS)
a lib with an alternate implementation
of the function checked with AC_CHECK_FUNC.
This works with ld.lld, but fails with
GNU ld, as configure then can't link it
(
https://sourceware.org/bugzilla/show_bug.cgi?id=32950
--- Comment #5 from Stas Sergeev ---
The problem is not in a makefile,
but in autotools itself. Consider
this example performed on a plain
"ncurses" source tree:
LDFLAGS="-Wl,--as-needed -lgcc" ./configure
configure passes because -lgcc
is
https://sourceware.org/bugzilla/show_bug.cgi?id=32950
--- Comment #8 from Stas Sergeev ---
Thanks, that seems to work.
Unfortunately pkg-config doesn't
have `--ldflags`, which is why
I haven't done that initially.
This is minor and can be ignored,
but maybe you have the plans to
add --ldflags to
101 - 118 of 118 matches
Mail list logo
