On 18 Nov 2002, Dax Kelson wrote:
>
> Linus, do you want compile failure reports like below?
Well, I certainly prefer not to, simply because you might as well send
them to linux-kernel and get it resolved that way by some random person
who just happens to be awake.
It looks like
- an ACPI configuration bug, where CONFIG_ACPI_SLEEP is allowed even
though CONFIG_SOFTWARE_SUSPEND is not enabled.
- keyboard.c and parts of VT not depending on CONFIG_INPUT
- device mapper using vcalloc() even though no such function has ever
existed in the kernel (it's vmalloc + memset()).
and I think the only way to fix these things is to publicly shame
maintainers into fixing the stuff they maintain.
Linus
---
> arch/i386/kernel/built-in.o: In function `do_suspend_lowlevel':
> arch/i386/kernel/built-in.o(.data+0x1304): undefined reference to `save_processor_state'
> arch/i386/kernel/built-in.o(.data+0x130a): undefined reference to `saved_context_esp'
....
> drivers/built-in.o: In function `kd_nosound':
> drivers/built-in.o(.text+0x420ea): undefined reference to `input_event'
> drivers/built-in.o(.text+0x4210c): undefined reference to `input_event'
....
> drivers/built-in.o: In function `kd_mksound':
> drivers/built-in.o(.text+0x421ab): undefined reference to `input_event'
> drivers/built-in.o: In function `kbd_bh':
> drivers/built-in.o(.text+0x42f07): undefined reference to `input_event'
> drivers/built-in.o(.text+0x42f2a): undefined reference to `input_event'
....
> drivers/built-in.o: In function `alloc_targets':
> drivers/built-in.o(.text+0x8d0a7): undefined reference to `vcalloc'
> drivers/built-in.o: In function `setup_indexes':
> drivers/built-in.o(.text+0x8dabb): undefined reference to `vcalloc'
On Mon, Nov 18, 2002 at 11:03:54PM -0800, Linus Torvalds wrote:
>
> On 18 Nov 2002, Dax Kelson wrote:
> >
> > Linus, do you want compile failure reports like below?
>
> Well, I certainly prefer not to, simply because you might as well send
> them to linux-kernel and get it resolved that way by some random person
> who just happens to be awake.
This sounds slightly tinderboxy in Mozilla slang, perhaps that is an idea?
I know Ben (I think?) has a .config that currently compiles with 2.5 but
also includes almost everything that is working right now and so quickly
exposes stuff that breaks.
Perhaps we could set up a tinderbox like thing to autmatically assign 'BK
blame'.
Regards,
bert
--
http://www.PowerDNS.com Versatile DNS Software & Services
http://lartc.org Linux Advanced Routing & Traffic Control HOWTO
On Tue, 19 Nov 2002, bert hubert wrote:
>
> I know Ben (I think?) has a .config that currently compiles with 2.5 but
> also includes almost everything that is working right now and so quickly
> exposes stuff that breaks.
On the other hand, we do need to start fixing the odd stuff too, not just
keep one "wide" config working. Usually the "make an arbitrary
'randconfig' always work" phase comes fairly late in the game, and I'm not
saying we should even try to aim for that right now, but on the other hand
we should probably clean up some of the unnecessary confusion with config
options where the problem sometimes simply is that the Kconfig files don't
contain the right dependencies, for example.
Yeah, considering how 2.5.48 doesn't compile with modules disabled, I'm
throwing stones in glass houses, but let's try to fix the stupid ones so
that people can see what the real problems are..
Linus