Re: [SeaBIOS] [PATCH v2] qemu: fast boot when linuxboot optionrom is used

[Stefano Garzarella]

On Wed, Nov 28, 2018 at 3:38 PM Kevin O'Connor  wrote:
> On Wed, Nov 28, 2018 at 12:30:42PM +0100, Stefano Garzarella wrote:
> > On Wed, Nov 28, 2018 at 3:12 AM Kevin O'Connor  wrote:
> > > On Mon, Nov 26, 2018 at 12:39:12PM +0100, Stefano Garzarella wrote:
> > > > Speed up the boot phase when qemu uses "linuxboot" optionrom
> > > > (qemu -kernel) and the boot-menu is not required.
> > > > Under these conditions we can skip the setup of devices and VGA,
> > > > because they will be initialized (if they are required) during
> > > > the Linux boot phase.
> > > >
> > > > Following the time measured between SeaBIOS entry point and
> > > > "linuxboot" entry point:
> > > >
> > > > Before this patch
> > > > qemu -kernel  | qemu -vga none -kernel
> > > > --------------+-----------------------
> > > > 53.5 msec     | 23.34 msec
> > > >
> > > > After this patch
> > > > qemu -kernel  | qemu -vga none -kernel
> > > > --------------+-----------------------
> > > > 12.82 msec    | 10.89 msec
> > > >
> > > > Note: For the measuring, we used the default configuration disabling
> > > > debug messages (CONFIG_DEBUG_LEVEL=0) and applying Stephen's patch:
> > > > "tpm: Check for TPM related ACPI tables before attempting hw"
> > > > Signed-off-by: Stefano Garzarella
> > > >
> > > > src/bootsplash.c  |  3 +++
> > > >  src/fw/paravirt.c | 10 ++++++++++
> > > >  src/fw/paravirt.h |  4 ++++
> > > >  src/optionroms.c  |  3 ++-
> > > >  src/post.c        |  3 +++
> > > >  5 files changed, 22 insertions(+), 1 deletion(-)
> > > >
> > > > diff --git a/src/bootsplash.c b/src/bootsplash.c
> > > > index 165c98d..0eda7f2 100644
> > > > --- a/src/bootsplash.c
> > > > +++ b/src/bootsplash.c
> > > > @@ -8,6 +8,7 @@
> > > >  #include "bregs.h" // struct bregs
> > > >  #include "config.h" // CONFIG_*
> > > >  #include "farptr.h" // FLATPTR_TO_SEG
> > > > +#include "fw/paravirt.h" // runningOnQEMUFastBoot
> > > >  #include "malloc.h" // free
> > > >  #include "output.h" // dprintf
> > > >  #include "romfile.h" // romfile_loadfile
> > > > @@ -39,6 +40,8 @@ call16_int10(struct bregs *br)
> > > >  void
> > > >  enable_vga_console(void)
> > > >  {
> > > > if(runningOnQEMUFastBoot())
> > > >    return;
> > > > dprintf(1, "Turning on vga text mode console\n");
> > > >    struct bregs br;
> > > >
> > > > diff --git a/src/fw/paravirt.c b/src/fw/paravirt.c
> > > > index 0770c47..9e6e618 100644
> > > > --- a/src/fw/paravirt.c
> > > > +++ b/src/fw/paravirt.c
> > > > @@ -621,4 +621,14 @@ void qemu_cfg_init(void)
> > > >      if (nogfx && !romfile_find("etc/sercon-port")
> > > >          && !romfile_find("vgaroms/sgabios.bin"))
> > > >          const_romfile_add_int("etc/sercon-port", PORT_SERIAL1);
> > > > /*
> > > > * Enable QEMU fast boot if there is "linuxboot" optionrom and
> > > > * the boot menu is not required.
> > > > */
> > > > if ((romfile_find("genroms/linuxboot_dma.bin")
> > > >    || romfile_find("genroms/linuxboot.bin"))
> > > >    && !romfile_loadint("etc/show-boot-menu", 1)) {
> > > >    PlatformRunningOn |= PF_QEMU_FB;
> > > > }
> > >
> > > I don't think we should hardcode special meanings to the names of
> > > bootable files.  If QEMU wants SeaBIOS to not perform some type of
> > > hardware init, then I think QEMU should explicitly request that from
> > > SeaBIOS (eg, a "etc/dont-run-hardware-init").
> >
> > I agree, it is cleaner. That was one of my doubt but, with an explicit
> > request, it works only with a new version of QEMU.
> > Do you think is acceptable?
>
> It would be good to understand what the source of the delay is.  It
> might be possible to avoid the delay without adding a new config
> setting.  For example, the patch above will prevent the vga rom from
> being run, but it should be possible to accomplish the same with
> "-device VGA,romfile=".  The other change this patch makes is to skip
> device_hardware_setup() - is it known which items in that function add
> to the delay?

Hi Kevin,
thanks for your advice.

I tried to understand what are the sources of the delay, and I found
that most of the time is spent because of QEMU default devices (eg
e1000 ROM PCI, mouse, keyboard, etc)
At this point, (without the patch above) I disabled the default QEMU
devices (-nodefaults) and I reached 12.5 ms of SeaBIOS boot time
(compared to 22 ms, where 4 ms is taken by the loading of e1000 ROM).

Other 2 ms is taken by enable_vga_console(), where simply comment out
the printf("SeaBIOS (version %s)\n", VERSION) I reached 10.7 ms of
boot time.
Do you think is acceptable to use dprintf(1, ...) instead of printf()?
Or maybe check if we found the VGA.

Note: All these tests are done with CONFIG_DEBUG_LEVEL=0, using
CONFIG_DEBUG_LEVEL=1 the boot time grows up to 24 ms, maybe we should
put CONFIG_DEBUG_LEVEL=0 in the SeaBIOS configuration used in QEMU.
> Also, what is the target boot time that is considered acceptable?

I'm using qboot as a reference (6/7 ms). It is very minimal, so 10 ms
I think is considered acceptable for SeaBIOS

Thanks,
Stefano
> Thanks,
> -Kevin--

Stefano Garzarella
Red Hat