Re: Bug#877735: busybox: m68k build broken due to "-Os" in CFLAGS

Re: Bug#877735: busybox: m68k build broken due to "-Os" in CFLAGS

[John Paul Adrian Glaubitz]

Hi!

On 10/05/2017 01:16 AM, John Paul Adrian Glaubitz wrote:
> Since busybox is very important to boot the system and get
> debian-installer working, I would like to ask to have the
> change from [1] reverted for m68k until we have fixed the
> potential bug in gcc which most likely mis-compiled busybox.

On second thought, I would actually recommended to revert this change
for all architectures. Size isn't so much a constraint anymore these
days, you reduce the binary by about 200k. I don't think this is worth
the risk of breaking something as fundemantal as busybox.

Adrian

-- 
  .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - glaubitz@debian.org
`. `'   Freie Universitaet Berlin - glaubitz@physik.fu-berlin.de
   `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913

Re: Bug#877735: busybox: m68k build broken due to "-Os" in CFLAGS

[Roger Shimizu]

On Thu, Oct 5, 2017 at 4:25 PM, John Paul Adrian Glaubitz
<glaubitz@physik.fu-berlin.de> wrote:
> Hi!
>
> On 10/05/2017 01:16 AM, John Paul Adrian Glaubitz wrote:
>>
>> Since busybox is very important to boot the system and get
>> debian-installer working, I would like to ask to have the
>> change from [1] reverted for m68k until we have fixed the
>> potential bug in gcc which most likely mis-compiled busybox.
>
>
> On second thought, I would actually recommended to revert this change
> for all architectures. Size isn't so much a constraint anymore these
> days, you reduce the binary by about 200k. I don't think this is worth
> the risk of breaking something as fundemantal as busybox.

Please exclude armel, which has size limitation on initrd of d-i [0].
Thanks!

[0] https://anonscm.debian.org/cgit/d-i/debian-installer.git/tree/build/boot/arm/qnap-flash-debian#n13

Cheers,
-- 
Roger Shimizu, GMT +9 Tokyo
PGP/GPG: 4096R/6C6ACD6417B3ACB1

Re: Bug#877735: busybox: m68k build broken due to "-Os" in CFLAGS

[John Paul Adrian Glaubitz]

On 10/05/2017 05:55 PM, Roger Shimizu wrote:
>> On second thought, I would actually recommended to revert this change
>> for all architectures. Size isn't so much a constraint anymore these
>> days, you reduce the binary by about 200k. I don't think this is worth
>> the risk of breaking something as fundemantal as busybox.
> 
> Please exclude armel, which has size limitation on initrd of d-i [0].
Hmm, ok. Is that currently actually a concern? If I'm seeing that correctly,
the "-Os" here only saved us around 200k. Does that already make a difference
on armel?

If yes, I'm also fine with something like

# On m68k, building with -Os leads to miscompiled code
ifeq (m68k,$(DEB_HOST_ARCH))
	CFLAGS := $(filter-out -Werror=format-security, ${CFLAGS})
else
	CFLAGS := $(filter-out -Werror=format-security -O2, ${CFLAGS})
endif

Adrian

-- 
  .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - glaubitz@debian.org
`. `'   Freie Universitaet Berlin - glaubitz@physik.fu-berlin.de
   `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913

Re: Bug#877735: busybox: m68k build broken due to "-Os" in CFLAGS

[Steve McIntyre]

On Thu, Oct 05, 2017 at 06:12:46PM +0200, John Paul Adrian Glaubitz wrote:
>On 10/05/2017 05:55 PM, Roger Shimizu wrote:
>> > On second thought, I would actually recommended to revert this change
>> > for all architectures. Size isn't so much a constraint anymore these
>> > days, you reduce the binary by about 200k. I don't think this is worth
>> > the risk of breaking something as fundemantal as busybox.
>> 
>> Please exclude armel, which has size limitation on initrd of d-i [0].
>Hmm, ok. Is that currently actually a concern? If I'm seeing that correctly,
>the "-Os" here only saved us around 200k. Does that already make a difference
>on armel?

Massively so, yes. Lots of the armel platforms that people care about
have very limited space for kernel and initramfs.

Re: Bug#877735: busybox: m68k build broken due to "-Os" in CFLAGS

[Christoph Biedl]

[-- Attachment #1: Type: text/plain, Size: 991 bytes --]

Steve McIntyre wrote...

> On Thu, Oct 05, 2017 at 06:12:46PM +0200, John Paul Adrian Glaubitz wrote:

> >Hmm, ok. Is that currently actually a concern? If I'm seeing that correctly,
> >the "-Os" here only saved us around 200k. Does that already make a difference
> >on armel?
>
> Massively so, yes. Lots of the armel platforms that people care about
> have very limited space for kernel and initramfs.

Indeed. When we (KiBi, Chris and I) discussed maintaining busybox at
DebConf, such size concerns were on the list of things to keep in mind.
For that reason we will enable more applets upon justified request only.

About that particular issue I'd really like to avoid a per-architecture
switch in busybox. It adds complexity and works around a problem
instead of solving it. So please take this to gcc first unless you've
already done so. And let us know if you get the impression this won't
get fixed soon-ish - breaking an architecture still beats packaging
principles.

    Christoph

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: Bug#877735: busybox: m68k build broken due to "-Os" in CFLAGS

[John Paul Adrian Glaubitz]

On 10/05/2017 11:49 PM, Christoph Biedl wrote:
> About that particular issue I'd really like to avoid a per-architecture
> switch in busybox. It adds complexity and works around a problem
> instead of solving it.

I wouldn't call that simple if-clause complexity.

> So please take this to gcc first unless you've
> already done so. And let us know if you get the impression this won't
> get fixed soon-ish - breaking an architecture still beats packaging
> principles.

gcc bugs aren't usually something that get fixed promptly, in particular
when talking about non-mainstream architectures like m68k.

I will mark busybox as "Not-For-Us" for the time being now because I
don't want to be chasing this package over this and keep building
it manually so that debian-installer doesn't break for us.

Adrian

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - glaubitz@debian.org
`. `'   Freie Universitaet Berlin - glaubitz@physik.fu-berlin.de
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913

Re: Bug#877735: busybox: m68k build broken due to "-Os" in CFLAGS

[Steve McIntyre]

On Thu, Oct 05, 2017 at 09:25:12AM +0200, John Paul Adrian Glaubitz wrote:
>Hi!
>
>On 10/05/2017 01:16 AM, John Paul Adrian Glaubitz wrote:
>> Since busybox is very important to boot the system and get
>> debian-installer working, I would like to ask to have the
>> change from [1] reverted for m68k until we have fixed the
>> potential bug in gcc which most likely mis-compiled busybox.
>
>On second thought, I would actually recommended to revert this change
>for all architectures. Size isn't so much a constraint anymore these
>days, you reduce the binary by about 200k. I don't think this is worth
>the risk of breaking something as fundemantal as busybox.

I'd be more worried about the compiler breakage, to be honest. -Os is
meant to work!

Re: Bug#877735: busybox: m68k build broken due to "-Os" in CFLAGS

[John Paul Adrian Glaubitz]

On 10/05/2017 06:38 PM, Steve McIntyre wrote:
>> On second thought, I would actually recommended to revert this change
>> for all architectures. Size isn't so much a constraint anymore these
>> days, you reduce the binary by about 200k. I don't think this is worth
>> the risk of breaking something as fundemantal as busybox.
> 
> I'd be more worried about the compiler breakage, to be honest. -Os is
> meant to work!

Yes, it is meant to work. But it is not as guaranteed to work as -O2
simply because the majority of packages are built with -O2. So, there
is some potential of things breaking.

Of course, these issues should be fixed and I will file a bug report
on the problem. I just meant that varying the optimization levels
is not something that you can expect to work in 100% of the cases.

Adrian

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - glaubitz@debian.org
`. `'   Freie Universitaet Berlin - glaubitz@physik.fu-berlin.de
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913