lord@xxxxxxx wrote:
I created new patches today.
You may want to download the cvs patch and generate a a fresh tree.
Save your .config from you previous tree, and use it to seed your tree.
> > On Fri, 21 Jul 2000 11:41:27 -0700,
> > Rawb <rawb@xxxxxxxxxxxx> wrote:
> > >make -C xfs
> > >make[2]: Entering directory `/usr.ext/src/linux-2.4-xfs/linux/fs/xfs'
> > >/usr/src/linux-2.4-xfs/linux/Rules.make:75: warning: overriding commands
> > >for target `xfs.o'
> > >Makefile:93: warning: ignoring old commands for target `xfs.o'
> >
> > I cannot reproduce your problem with the current 2.4.0-test1-xfs
> > source. I am running from the SGI p_tools tree, not CVS, I wonder if
> > CVS is out of sync. It is puzzling that the Rules.make:75 and
> > fs/xfs/Makefile:93 lines have nothing to do with target xfs.o.
> >
> > Rawb, please mail me (not the list) your .config, Rules.make, top level
> > Makefile, fs/xfs/Makefile and the output from "md5sum fs/xfs/*".
>
> The key factor here is that there are no stubs at the start of the list of .o
> files:
>
> ld -m elf_i386 -r -o xfs.o xfs_alloc.o xfs_alloc_btree.o xfs_attr.o
> xfs_attr_leaf.o xfs_bit.o xfs_bmap.o xfs_bmap_btree.o xfs_btree.o
> xfs_buf_item.o xfs_da_btree.o xfs_dir.o xfs_dir2.o xfs_dir2_block.o
>
> where the source of the makefile says:
>
> ifeq ($(CONFIG_XFS_RT),y)
> XFS_RT_OBJS=xfs_rtalloc.o xfs_rtbit.o
> else
> XFS_RT_OBJS=xfsrtstubs.o
> endif
> ifeq ($(CONFIG_XFS_GRIO),y)
> XFS_GRIO_OBJS=xfs_grio.o
> else
> XFS_GRIO_OBJS=
> endif
> ifeq ($(CONFIG_XFS_DMAPI),y)
> XFS_DMAPI_OBJS=dmapi/dmapi.o
> else
> XFS_DMAPI_OBJS=xfsdmapistubs.o
> endif
>
> XFS_OBJS := $(XFS_RT_OBJS) \
> $(XFS_GRIO_OBJS) \
> $(XFS_DMAPI_OBJS) \
> xfs_alloc.o \
> xfs_alloc_btree.o \
>
> This has happened before, when the grio and realtime flags got taken out of
> the config options. There is some cached state in the tree which gets confused
> when these flags change I think. Saving the .config filesomewhere, doing
> a make mrproper, reinstalling the .config and doing a make oldconfig might
> fix it.
>
> It could have something to do with the include/config file state beiong
> messed up.
>
> Steve (temporarily in ottawa)
>
> options.
--
Russell Cattelan
cattelan@xxxxxxxxxxx
|