This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: tst-mallocstate.c failure
- From: Florian Weimer <fweimer at redhat dot com>
- To: Steve Ellcey <sellcey at caviumnetworks dot com>
- Cc: libc-alpha at sourceware dot org
- Date: Wed, 14 Dec 2016 22:36:37 +0100
- Subject: Re: tst-mallocstate.c failure
- Authentication-results: sourceware.org; auth=none
- References: <1481750003.3445.2.camel@caviumnetworks.com>
On 12/14/2016 10:13 PM, Steve Ellcey wrote:
Florian,
The tst-mallocstate test you created is failing for me with my ILP32 aarch64
changes. This is a new ABI for 2.25 so I am not sure how to
handle malloc_get_state. If I understand things correctly, malloc_get_state
has been removed from glibc and so should not appear in my ld.abilist,
is that right?
It should not appear in any abilist file.
When I did have it I get an error with check-abi-ld:
@@ -1235 +1233,0 @@ GLIBC_2.25 malloc F
-GLIBC_2.25 malloc_get_state F
But taking malloc_get_state out of aarch64.abilist didn't help
the actual tst-malloc-state failure which dies with:
/home/ubuntu/sellcey/glibc-patched/src/glibc/malloc/tst-mallocstate.c:493: undefined reference to `malloc_get_state@GLIBC_2.25'
/home/ubuntu/sellcey/glibc-patched/obj-glibc32/malloc/tst-mallocstate.o: In function `init_heap':
/home/ubuntu/sellcey/glibc-patched/src/glibc/malloc/tst-mallocstate.c:315: undefined reference to `malloc_set_state@GLIBC_2.25'
Do I need to create a dummy version of this routine somewhere? Or should
this test not be run since I don't have the function that it is trying to
test. I am not sure what needs to be done to resolve this test failure.
The symbol does not exist anymore, so this is expected. I didn't expect
a new architecture so soon. It's a pity that you'll still inherit loads
of legacy symbols we are about to remove (probably in 2.26).
We need to disable the test somehow. This is the first test for a
compatibility symbol, so this problem is quite new.
I think the proper solution is to generate something like abi-versions.h
for use in makefiles, and use that to make the existence of the test
conditional in the makefile. Or does anyone have a better idea?
Thanks,
Florian