This is the mail archive of the
libc-alpha@sourceware.cygnus.com
mailing list for the glibc project.
Re: Stuff dragged in by crt1 (was Re: Announce: initrd-tftp 0.1)
- To: Wolfram Gloger <Wolfram dot Gloger at dent dot med dot uni-muenchen dot de>
- Subject: Re: Stuff dragged in by crt1 (was Re: Announce: initrd-tftp 0.1)
- From: Philip Blundell <pb at labs dot futuretv dot com>
- Date: Tue, 11 Jan 2000 16:23:30 +0000
- cc: zack at wolery dot cumb dot org, libc-alpha at sourceware dot cygnus dot com
- References: <200001111448.PAA64253@max.zk-i.med.uni-muenchen.de>
>Well, I think MALLOC_CHECK_ is really immensely useful :), so I won't
>scrap it. A move of all the malloc stuff that uses stdio into a
>preload library would certainly be OK with me, but then MALLOC_CHECK_
>wouldn't work with statically linked executables, which are the topic
>at hand.
You could make a libmalloc_check.a, along the same lines as the preload
library, for them to link against. But quite honestly nobody cares about
debugging static binaries - the only time you use them is when you are
building an image to go in a ROM or on a floppy disk, by which time your code
should already be working.
>Anyway, I am considering putting it on my TODO to at least move all
>the malloc stuff that uses stdio into a separate source file (the
>single malloc source file has become way too large in any case). Then
>when linking statically stdio wouldn't be pulled in unless you would
>call __malloc_check_init or malloc_stats directly.
That would be great.
>I'm not sure yet whether all this is worth it. Comments?
Feedback from embedded people suggests that it is. I think the only real
alternative is for them to make their own library containing a malloc without
the checking stuff.
p.