Next version

[Kevin O'Connor]

Hi,

I'd like to make the next release of SeaBIOS.  I plan to make the
version number be v1.6.3 (instead of v0.6.3).

There are a few ACPI patches pending.  However, I think it would be
best to tag the current code, and then add the ACPI code on top of it.

If anyone knows of any bugs, patches, or issues that need to be
addressed prior to making a release, please let me know.

Thanks,
-Kevin

[Paolo Bonzini]

On 09/23/2011 06:28 AM, Kevin O'Connor wrote:
> I'd like to make the next release of SeaBIOS.  I plan to make the
> version number be v1.6.3 (instead of v0.6.3).
>
> There are a few ACPI patches pending.  However, I think it would be
> best to tag the current code, and then add the ACPI code on top of it.
>
> If anyone knows of any bugs, patches, or issues that need to be
> addressed prior to making a release, please let me know.

Two-pass PCI allocation caused a regression that was recently reported 
in qemu-devel.

Paolo

[Ian Campbell]

On Fri, 2011-09-23 at 00:28 -0400, Kevin O'Connor wrote:
> Hi,
>
> I'd like to make the next release of SeaBIOS.  I plan to make the
> version number be v1.6.3 (instead of v0.6.3).
>
> There are a few ACPI patches pending.  However, I think it would be
> best to tag the current code, and then add the ACPI code on top of it.
>
> If anyone knows of any bugs, patches, or issues that need to be
> addressed prior to making a release, please let me know.

I've just tried to build and got:

  Linking out/rom.o
`.discard.var16.src/stacks.c.188' referenced in section `.text.finish_preempt.1580' of out/code32flat.o: defined in discarded section `.discard.var16.src/stacks.c.188' of out/code32flat.o
`.discard.var16.src/stacks.c.188' referenced in section `.text.pmm_malloc.1805' of out/code32flat.o: defined in discarded section `.discard.var16.src/stacks.c.188' of out/code32flat.o
`.discard.var16.src/stacks.c.188' referenced in section `.text.pmm_malloc.1805' of out/code32flat.o: defined in discarded section `.discard.var16.src/stacks.c.188' of out/code32flat.o
`.discard.var16.src/stacks.c.188' referenced in section `.text.init.src/pmm.c.561' of out/code32flat.o: defined in discarded section `.discard.var16.src/stacks.c.188' of out/code32flat.o
`.discard.var16.src/stacks.c.188' referenced in section `.text.init.src/pmm.c.561' of out/code32flat.o: defined in discarded section `.discard.var16.src/stacks.c.188' of out/code32flat.o
`.discard.var16.src/stacks.c.188' referenced in section `.text.init.src/pmm.c.561' of out/code32flat.o: defined in discarded section `.discard.var16.src/stacks.c.188' of out/code32flat.o
`.discard.var16.src/stacks.c.188' referenced in section `.text.ehci_note_port.41146' of out/code32flat.o: defined in discarded section `.discard.var16.src/stacks.c.188' of out/code32flat.o
`.discard.var16.src/stacks.c.188' referenced in section `.text.ehci_note_port.41146' of out/code32flat.o: defined in discarded section `.discard.var16.src/stacks.c.188' of out/code32flat.o
`.discard.var16.src/stacks.c.188' referenced in section `.text.__callrom.68478' of out/code32flat.o: defined in discarded section `.discard.var16.src/stacks.c.188' of out/code32flat.o
`.discard.var16.src/stacks.c.188' referenced in section `.text.maininit.53837' of out/code32flat.o: defined in discarded section `.discard.var16.src/stacks.c.188' of out/code32flat.o

Which I bisected to:

commit 533b6286cbc5a01e8aaafc2a8393a7d109f81958
Author: Kevin O'Connor <[e-mail]>
Date:   Sat Jul 16 13:13:12 2011 -0400

    Run option rom visible PMM code in 32bit mode instead of 16bit mode.
    
    Use call32() to jump into handle_pmm().  This reduces the amount of
    16bit code needed.
    
    Signed-off-by: Kevin O'Connor <[e-mail]>

I tried sprinkling some VISIBLE* stuff around the place but it quickly
became pretty clear I didn't know what I was doing ;-)

Turning off CONFIG_PMM "fixed" it, original (failing) .config is
attached.

Thanks,
Ian.

-- 
Ian Campbell
Current Noise: Crowbar - Lifesblood

You are a very redundant person, that's what kind of person you are.

[Kevin O'Connor]

On Thu, Sep 29, 2011 at 02:00:11PM +0100, Ian Campbell wrote:
> On Fri, 2011-09-23 at 00:28 -0400, Kevin O'Connor wrote:
> > Hi,
> >
> > I'd like to make the next release of SeaBIOS.  I plan to make the
> > version number be v1.6.3 (instead of v0.6.3).
> >
> > There are a few ACPI patches pending.  However, I think it would be
> > best to tag the current code, and then add the ACPI code on top of it.
> >
> > If anyone knows of any bugs, patches, or issues that need to be
> > addressed prior to making a release, please let me know.
>
> I've just tried to build and got:
>
> Linking out/rom.o
> `.discard.var16.src/stacks.c.188' referenced in section `.text.finish_preempt.1580' of out/code32flat.o: defined in discarded section `.discard.var16.src/stacks.c.188' of out/code32flat.o

Ughh - this is likely an issue with gcc's -combine option.  What OS
and/or gcc/ld version are you using?

-Kevin

[Ian Campbell]

On Thu, 2011-09-29 at 19:36 -0400, Kevin O'Connor wrote:
> On Thu, Sep 29, 2011 at 02:00:11PM +0100, Ian Campbell wrote:
> > On Fri, 2011-09-23 at 00:28 -0400, Kevin O'Connor wrote:
> > > Hi,
> > >
> > > I'd like to make the next release of SeaBIOS.  I plan to make the
> > > version number be v1.6.3 (instead of v0.6.3).
> > >
> > > There are a few ACPI patches pending.  However, I think it would be
> > > best to tag the current code, and then add the ACPI code on top of it.
> > >
> > > If anyone knows of any bugs, patches, or issues that need to be
> > > addressed prior to making a release, please let me know.
> >
> > I've just tried to build and got:
> >
> > Linking out/rom.o
> > `.discard.var16.src/stacks.c.188' referenced in section `.text.finish_preempt.1580' of out/code32flat.o: defined in discarded section `.discard.var16.src/stacks.c.188' of out/code32flat.o
>
> Ughh - this is likely an issue with gcc's -combine option.  What OS
> and/or gcc/ld version are you using?

It is Debian Squeeze.

$ gcc --version
gcc (Debian 4.4.5-8) 4.4.5
$ ld --version
GNU ld (GNU Binutils for Debian) 2.20.1-system.20100303

Ian.

-- 
Ian Campbell


Another good night not to sleep in a eucalyptus tree.

[Bjørn Mork]

Ian Campbell <[e-mail]> writes:
> On Thu, 2011-09-29 at 19:36 -0400, Kevin O'Connor wrote:
> > > I've just tried to build and got:
> > >
> > > Linking out/rom.o
> > > `.discard.var16.src/stacks.c.188' referenced in section `.text.finish_preempt.1580' of out/code32flat.o: defined in discarded section `.discard.var16.src/stacks.c.188' of out/code32flat.o
> >
> > Ughh - this is likely an issue with gcc's -combine option.  What OS
> > and/or gcc/ld version are you using?
>
> It is Debian Squeeze.
>
> $ gcc --version
> gcc (Debian 4.4.5-8) 4.4.5
> $ ld --version
> GNU ld (GNU Binutils for Debian) 2.20.1-system.20100303

Then I don't think that can be the (only) cause.  I also use  Debian
Squeeze with those exact versions, and do not experience the problem:

bjorn@canardo:/usr/local/src/git/seabios$ git branch --contains 533b6286c
* master
bjorn@canardo:/usr/local/src/git/seabios$ make distclean
bjorn@canardo:/usr/local/src/git/seabios$ make
  Build default config
#
# configuration written to /usr/local/src/git/seabios/.config
#
  Build Kconfig config file
  Compiling whole program out/ccode.16.s
  Compiling to assembler out/asm-offsets.s
  Generating offset file out/asm-offsets.h
  Compiling (16bit) out/code16.o
  Compiling whole program out/ccode32flat.o
  Compiling whole program out/code32seg.o
  Building ld scripts (version "pre-0.6.3-20110930_094155-canardo")
Fixed space: 0xe05b-0x10000  total: 8101  slack: 11  Percent slack: 0.1%
16bit size:           40480
32bit segmented size: 1436
32bit flat size:      13716
32bit flat init size: 61968
  Linking out/rom16.o
  Stripping out/rom16.strip.o
  Linking out/rom32seg.o
  Stripping out/rom32seg.strip.o
  Linking out/rom.o
  Prepping out/bios.bin
Total size: 119984  Fixed: 55644  Free: 11088 (used 91.5% of 128KiB rom)



Bjørn

[Ian Campbell]

On Fri, 2011-09-30 at 09:44 +0200, Bjørn Mork wrote:
> Ian Campbell <[e-mail]> writes:
> > On Thu, 2011-09-29 at 19:36 -0400, Kevin O'Connor wrote:
> > > > I've just tried to build and got:
> > > >
> > > > Linking out/rom.o
> > > > `.discard.var16.src/stacks.c.188' referenced in section `.text.finish_preempt.1580' of out/code32flat.o: defined in discarded section `.discard.var16.src/stacks.c.188' of out/code32flat.o
> > >
> > > Ughh - this is likely an issue with gcc's -combine option.  What OS
> > > and/or gcc/ld version are you using?
> >
> > It is Debian Squeeze.
> >
> > $ gcc --version
> > gcc (Debian 4.4.5-8) 4.4.5
> > $ ld --version
> > GNU ld (GNU Binutils for Debian) 2.20.1-system.20100303
>
> Then I don't think that can be the (only) cause.  I also use  Debian
> Squeeze with those exact versions, and do not experience the problem:
> bjorn@canardo:/usr/local/src/git/seabios$ git branch --contains 533b6286c
> * master
> bjorn@canardo:/usr/local/src/git/seabios$ make distclean
> bjorn@canardo:/usr/local/src/git/seabios$ make
>   Build default config

This works for me too. Bisecting between my .config and the default it
appears as if just enabling CONFIG_THREAD_OPTIONROMS over the defaults
causes the problem.

Ian.

-- 
Ian Campbell
Current Noise: Candlemass - Crystal Ball

Alas, I am dying beyond my means.
		-- Oscar Wilde [as he sipped champagne on his deathbed]

[Bjørn Mork]

Ian Campbell <[e-mail]> writes:
> This works for me too. Bisecting between my .config and the default it
> appears as if just enabling CONFIG_THREAD_OPTIONROMS over the defaults
> causes the problem.

Right.  Yes, I can confirm that causes the problem for me as well.


Bjørn

[Kevin O'Connor]

On Fri, Sep 30, 2011 at 10:05:54AM +0100, Ian Campbell wrote:
> On Fri, 2011-09-30 at 09:44 +0200, Bjørn Mork wrote:
> > Ian Campbell <[e-mail]> writes:
> > > On Thu, 2011-09-29 at 19:36 -0400, Kevin O'Connor wrote:
> > > > > I've just tried to build and got:
> > > > >
> > > > > Linking out/rom.o
> > > > > `.discard.var16.src/stacks.c.188' referenced in section `.text.finish_preempt.1580' of out/code32flat.o: defined in discarded section `.discard.var16.src/stacks.c.188' of out/code32flat.o
> > > >
> > > > Ughh - this is likely an issue with gcc's -combine option.  What OS
> > > > and/or gcc/ld version are you using?
> > >
> > > It is Debian Squeeze.
> > >
> > > $ gcc --version
> > > gcc (Debian 4.4.5-8) 4.4.5
> > > $ ld --version
> > > GNU ld (GNU Binutils for Debian) 2.20.1-system.20100303
> >
> > Then I don't think that can be the (only) cause.  I also use  Debian
> > Squeeze with those exact versions, and do not experience the problem:
> > bjorn@canardo:/usr/local/src/git/seabios$ git branch --contains 533b6286c
> > * master
> > bjorn@canardo:/usr/local/src/git/seabios$ make distclean
> > bjorn@canardo:/usr/local/src/git/seabios$ make
> >   Build default config
>
> This works for me too. Bisecting between my .config and the default it
> appears as if just enabling CONFIG_THREAD_OPTIONROMS over the defaults
> causes the problem.

Can you check if "make clean; make COMPSTRAT=1" builds successfully?

-Kevin

[Kevin O'Connor]

On Fri, Sep 30, 2011 at 07:17:57AM -0400, Kevin O'Connor wrote:
> On Fri, Sep 30, 2011 at 10:05:54AM +0100, Ian Campbell wrote:
> > This works for me too. Bisecting between my .config and the default it
> > appears as if just enabling CONFIG_THREAD_OPTIONROMS over the defaults
> > causes the problem.
>
> Can you check if "make clean; make COMPSTRAT=1" builds successfully?

Looking more closely, I'm pretty sure this is the compiler bug
described in tools/test-gcc.sh:

# Also, on several compilers, -combine fails if code is emitted with a
# reference to an extern variable that is later found to be externally
# visible - the compiler does not mark those variables as global.
# This is being worked around by ordering the compile objects to avoid
# this case.

as such, I think the patch below should fix it.

After the release, I think SeaBIOS should just drop support for
"-combine" - gcc 4.6 no longer supports it and -flto is the preferred
way forward.

-Kevin


--- a/Makefile
+++ b/Makefile
@@ -11,7 +11,7 @@ VERSION=pre-0.6.3-$(shell date +"%Y%m%d_%H%M%S")-$(shell hostname)
 OUT=out/
 
 # Source files
-SRCBOTH=misc.c pmm.c stacks.c output.c util.c block.c floppy.c ata.c mouse.c \
+SRCBOTH=misc.c stacks.c pmm.c output.c util.c block.c floppy.c ata.c mouse.c \
         kbd.c pci.c serial.c clock.c pic.c cdrom.c ps2port.c smp.c resume.c \
         pnpbios.c pirtable.c vgahooks.c ramdisk.c pcibios.c blockcmd.c \
         usb.c usb-uhci.c usb-ohci.c usb-ehci.c usb-hid.c usb-msc.c \

[Ian Campbell]

On Fri, 2011-09-30 at 07:31 -0400, Kevin O'Connor wrote:
> On Fri, Sep 30, 2011 at 07:17:57AM -0400, Kevin O'Connor wrote:
> > On Fri, Sep 30, 2011 at 10:05:54AM +0100, Ian Campbell wrote:
> > > This works for me too. Bisecting between my .config and the default it
> > > appears as if just enabling CONFIG_THREAD_OPTIONROMS over the defaults
> > > causes the problem.
> >
> > Can you check if "make clean; make COMPSTRAT=1" builds successfully?

It does. FYI without the override COMPSTRAT=0.
> Looking more closely, I'm pretty sure this is the compiler bug
> described in tools/test-gcc.sh:
>
> # Also, on several compilers, -combine fails if code is emitted with a
> # reference to an extern variable that is later found to be externally
> # visible - the compiler does not mark those variables as global.
> # This is being worked around by ordering the compile objects to avoid
> # this case.
>
> as such, I think the patch below should fix it.

Which it does. Thanks.

Ian.

-- 
Ian Campbell
Current Noise: Red Hot Chili Peppers - Sexy Mexican Maid

Friction is a drag.