svn commit: r335629 - in head: share/man/man4 sys/dev/vt/hw/vga

[Rodney W. Grimes]

> On Tue, Jun 26, 2018 at 10:18:39AM -0600, Ian Lepore wrote:
> > On Tue, 2018-06-26 at 08:55 -0700, Rodney W. Grimes wrote:
> > > > 
> > > > On Tue, Jun 26, 2018 at 05:21:27AM +0000, Alexey Dokuchaev wrote:
> > > > > 
> > > > > On Mon, Jun 25, 2018 at 08:43:51AM -0700, Rodney W. Grimes wrote:
> > > > > > 
> > > > > > > 
> > > > > > > New Revision: 335629
> > > > > > > URL: https://svnweb.freebsd.org/changeset/base/335629
> > > > > > > 
> > > > > > > Log:
> > > > > > > ? vt: add option to ignore NO_VGA flag in ACPI
> > > > > > > ??
> > > > > > > ? To workaround buggy firmware that sets this flag when
> > > > > > > there's actually
> > > > > > > ? a VGA present.
> > > > > > > ??
> > > > > > > ? Reported and tested by:	Yasuhiro KIMURA <yasu at utahim
> > > > > > > e.org>
> > > > > > > ? Sponsored by:		Citrix Systems R&D
> > > > > > > ? Reviewed by:		kib
> > > > > > > ? Differential revision:	https://reviews.freebsd.org/D
> > > > > > > 16003
> > > > > > It is generally best to avoid double negatives,
> > > > > > couldnt this of been better named? (hw.vga.acpi_force_vga)
> > > > > Yes please; I get constantly confused when calculating negatives
> > > > > and
> > > > > often get them wrong.
> > > > This is specifically done to workaround a firmware bug where some
> > > > buggy firmwares set the NO_VGA flag in ACPI.
> > > We are not conflicted about working around the buggy ACPI.
> > > 
> > > > 
> > > > So the option does
> > > > exactly what the name says, it ignores the NO_VGA flag in ACPI. IMO
> > > > acpi_force_vga is not as descriptive as the current name.
> > > Interestingly that is the text you use to describe it in the man
> > > page, so it seems as if it is good for the description, but not
> > > good for the name of the flag itself?
> > > 
> > > .It Va hw.vga.acpi_ignore_no_vga
> > > Set to 1 to force the usage of the VGA driver regardless of whether
> > > ACPI IAPC_BOOT_ARCH signals no VGA support.
> > > Can be used to workaround firmware bugs in the ACPI tables.
> > > 
> > > This does not mention the ACPI table entry being over ridden,
> > > 	 if (flags & ACPI_FADT_NO_VGA)
> > > 
> > > Further digging I believe you have placed this in the wrong
> > > part of the hierarchy.??You put it in hw.vga, and it really
> > > should be in hw.acpi.
> > > Maybe hw.acpi.bootflags.ignore.no_vga.
> > > 
> > > Are there any other bootflags we may want to ignore?
> > > 
> > > Regards,
> > 
> > 
> > There is ACPI_FADT_NO_CMOS_RTC, with an associated tunable
> > hw.atrtc.enabled (default is -1), which can be:
Shouldnt this really be in hw.acpi.....

> > 
> > ? -1 - enabled unless acpi says ACPI_FADT_NO_CMOS_RTC
> > ? ?0 - unconditionally disabled
> > ? ?1 - unconditionally enabled
> > 
> > The idea was that if RTC is provided by EFI runtime services, this flag
> > would indicate that old-school CMOS RTC drivers should not be used.
> > ?But, predictably, it turns out there are bioses that set this flag
> > even when booting in legacy non-efi mode, leading to a need to ignore
> > the flag and force use of the old driver.
> 
> I could take this approach for vt and add hw.vt.enabled = { -1, 0, 1 }
> if that seems better.

Lets not scatter more ACPI stuff into non ACPI parts
of the sysctl oid tree.
If you do go this route it should really be in hw.acpi.*


-- 
Rod Grimes                                                 rgrimes at freebsd.org