[PATCH] newlib: libc: merge build up a directory

[Mike Frysinger]

On 14 Mar 2022 11:58, Corinna Vinschen wrote:
> On Mar 13 22:48, Mike Frysinger wrote:
> > On 01 Mar 2022 20:39, Mike Frysinger wrote:
> > > --- a/newlib/libc/stdio/Makefile.inc
> > > +++ b/newlib/libc/stdio/Makefile.inc
> > > ...
> > > +	%D%/vfiwscanf.c \
> > > +	%D%/vfwscanf.c
> > > +	%D%/vprintf.c \
> > > ...
> > 
> > found a missing \ here fwiw
> 
> The patch applies and I added that backslash.  Then I removed all
> Makefile.in files from newlib and called `autoreconf', which ran
> through without any warning or error.
> 
> However, clean building Cygwin with parallel make fails reproducably:
> 
>   $ $src/configure
>   $ make -j42
>   make[3]: Entering directory '$bin/x86_64-pc-cygwin/newlib'
>     GEN      stmp-targ-include
>   make[3]: *** No rule to make target 'fdopen64.c', needed by 'libc_a-fdopen64.o'.  Stop.
>   make[3]: *** Waiting for unfinished jobs....
> 
> So I tried with make -j1 and it still fails the same way.
> 
> This happens for both, i686 as well as x86_64 Cygwin.

my cygwin builds broke a while ago and i don't know why.  although i
just tried them again, and they've unbroke themselves.  who knows.

at any rate, looks like stdio64 is annoying in that cygwin is the only
target that tests it :(.  so that's why all my other ~30 targets are
passing.  i'll have a v2 up with all the issues fixed, and i'll do a
deep dive on the generated objects/libs.  i was holding off on doing
that until it sounded like people were OK with the general direction
of the code since it takes so long (both in terms of generating the
data, and in me digging through it).  i think we're at that point now.
-mike
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <https://sourceware.org/pipermail/newlib/attachments/20220314/daa006d9/attachment.sig>