Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.debian.kernel > #57964 > unrolled thread

Bug#864368: Expand kernel serial port configuration

Started byDave Hansen <dave.hansen@intel.com>
First post2017-06-07 18:50 +0200
Last post2017-06-07 19:20 +0200
Articles 5 — 3 participants

Back to article view | Back to linux.debian.kernel


Contents

  Bug#864368: Expand kernel serial port configuration Dave Hansen <dave.hansen@intel.com> - 2017-06-07 18:50 +0200
    Bug#864368: Expand kernel serial port configuration Ben Hutchings <ben@decadent.org.uk> - 2017-06-07 19:20 +0200
      Bug#864368: Expand kernel serial port configuration Ben Hutchings <ben@decadent.org.uk> - 2017-06-07 22:20 +0200
        Bug#864368: Expand kernel serial port configuration Dave Hansen <dave.hansen@intel.com> - 2017-06-07 22:40 +0200
    Processed: Re: Bug#864368: Expand kernel serial port configuration owner@bugs.debian.org (Debian Bug Tracking System) - 2017-06-07 19:20 +0200

#57964 — Bug#864368: Expand kernel serial port configuration

FromDave Hansen <dave.hansen@intel.com>
Date2017-06-07 18:50 +0200
SubjectBug#864368: Expand kernel serial port configuration
Message-ID<tPMjD-G2-3@gated-at.bofh.it>
Package: linux-image-4.9.0-3-amd64
Severity: medium

When my system boots the Debian installer, it appears to hang at a blank
screen.  There are two reasons for this, and two kernel configuration
parameters need to be modified:

CONFIG_SERIAL_8250_RUNTIME_UARTS=16
CONFIG_SERIAL_8250_MID=y

There are two things that cause the apparent hang:

One, the system has only has PCI serial ports.  The traditional, legacy
serial ports (at the low I/O ports like 0x3F8) are either not present or
not connected.  The kernel is only configured for 4 runtime serial ports
 (CONFIG_SERIAL_8250_RUNTIME_UARTS=4).  But, all 4 of those ports are
used for the legacy ports leaving no room for the PCI serial port.

Please raise:

	CONFIG_SERIAL_8250_RUNTIME_UARTS

to something larger, perhaps 16 or 8, such as:

	CONFIG_SERIAL_8250_RUNTIME_UARTS=16

The second issue is that the "mid" serial driver is not configured in
this package.  Please add the "8250_MID" driver by default and have it
built in (as opposed to being as a module).  Some systems which have
serial ports driven by this driver have *only* these ports.  They have
no other output devices.  Having it built in is the only way to diagnose
and fix boot issues that occur before modules start being loaded.

The necessary option is:

	CONFIG_SERIAL_8250_MID=y

While the 4.9 kernel is the most important one to see these options
changed, it would also be handy to have these options added for other
packages.  Should I submit separate bugs from those?

[toc] | [next] | [standalone]


#57965

FromBen Hutchings <ben@decadent.org.uk>
Date2017-06-07 19:20 +0200
Message-ID<tPMMG-14L-3@gated-at.bofh.it>
In reply to#57964

[Multipart message — attachments visible in raw view] — view raw

Control: reassign -1 src:linux 4.9.30-1
Control: severity -1 important

On Wed, 2017-06-07 at 09:38 -0700, Dave Hansen wrote:
[...]
> Please raise:
> 
> 	CONFIG_SERIAL_8250_RUNTIME_UARTS
> 
> to something larger, perhaps 16 or 8, such as:
> 
> 	CONFIG_SERIAL_8250_RUNTIME_UARTS=16

That seems reasonable.

> The second issue is that the "mid" serial driver is not configured in
> this package.  Please add the "8250_MID" driver by default and have it
> built in (as opposed to being as a module).  Some systems which have
> serial ports driven by this driver have *only* these ports.  They have
> no other output devices.  Having it built in is the only way to diagnose
> and fix boot issues that occur before modules start being loaded.
> 
> The necessary option is:
> 
> 	CONFIG_SERIAL_8250_MID=y

Sorry, I disabled this one for 4.4 because that was then the default. 
It seems reasonable to enable it for x86 kernels if they will actually
run on the relevant SoCs.

> While the 4.9 kernel is the most important one to see these options
> changed, it would also be handy to have these options added for other
> packages.  Should I submit separate bugs from those?

No, I'm reassigning it to the source package instead and we can mark it
found in the relevant versions.

Ben.

-- 
Ben Hutchings
If the facts do not conform to your theory, they must be disposed of.

[toc] | [prev] | [next] | [standalone]


#57968

FromBen Hutchings <ben@decadent.org.uk>
Date2017-06-07 22:20 +0200
Message-ID<tPPAS-2SK-9@gated-at.bofh.it>
In reply to#57965

[Multipart message — attachments visible in raw view] — view raw

On Wed, 2017-06-07 at 18:10 +0100, Ben Hutchings wrote:
> Control: reassign -1 src:linux 4.9.30-1
> Control: severity -1 important
> 
> On Wed, 2017-06-07 at 09:38 -0700, Dave Hansen wrote:
> [...]
> > Please raise:
> > 
> > 	CONFIG_SERIAL_8250_RUNTIME_UARTS
> > 
> > to something larger, perhaps 16 or 8, such as:
> > 
> > 	CONFIG_SERIAL_8250_RUNTIME_UARTS=16
> 
> That seems reasonable.
[...]

Actually, having looked further at what this does, I'm unconvinced
about changing it.  It can always be overridden through the
'8250.nr_uarts' kernel parameter where needed.

Ben.

-- 
Ben Hutchings
If the facts do not conform to your theory, they must be disposed of.

[toc] | [prev] | [next] | [standalone]


#57969

FromDave Hansen <dave.hansen@intel.com>
Date2017-06-07 22:40 +0200
Message-ID<tPPUe-30Y-11@gated-at.bofh.it>
In reply to#57968
On 06/07/2017 01:14 PM, Ben Hutchings wrote:
> On Wed, 2017-06-07 at 18:10 +0100, Ben Hutchings wrote:
>>> 	CONFIG_SERIAL_8250_RUNTIME_UARTS=16
>> That seems reasonable.
> [...]
> 
> Actually, having looked further at what this does, I'm unconvinced
> about changing it.  It can always be overridden through the
> '8250.nr_uarts' kernel parameter where needed.

On my system, this meant that I could not use the serial console to do
an install.  Doing console=ttyS4 on the command-line didn't work because
the kernel didn't support anything after ttyS3.

I guess we could also teach folks to add nr_uarts=8 (or whatever) to
their kernel command-line, which also works around this.  But, it would
be handy to have it work by default.

I don't think there's really any overhead to raising 'nr_uarts', either.
 It makes some (unimportant) loops longer, but I don't think there's any
real downsides.

FWIW, a lot of distros make it pretty large.  My Ubuntu desktop has it
set to 48 and Fedora evidently sets it to 32.

[toc] | [prev] | [next] | [standalone]


#57966 — Processed: Re: Bug#864368: Expand kernel serial port configuration

Fromowner@bugs.debian.org (Debian Bug Tracking System)
Date2017-06-07 19:20 +0200
SubjectProcessed: Re: Bug#864368: Expand kernel serial port configuration
Message-ID<tPMMG-14L-13@gated-at.bofh.it>
In reply to#57964
Processing control commands:

> reassign -1 src:linux 4.9.30-1
Bug #864368 [linux-image-4.9.0-3-amd64] Expand kernel serial port configuration
Bug reassigned from package 'linux-image-4.9.0-3-amd64' to 'src:linux'.
Ignoring request to alter found versions of bug #864368 to the same values previously set
Ignoring request to alter fixed versions of bug #864368 to the same values previously set
Bug #864368 [src:linux] Expand kernel serial port configuration
Marked as found in versions linux/4.9.30-1.
> severity -1 important
Bug #864368 [src:linux] Expand kernel serial port configuration
Severity set to 'important' from 'normal'

-- 
864368: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=864368
Debian Bug Tracking System
Contact owner@bugs.debian.org with problems

[toc] | [prev] | [standalone]


Back to top | Article view | linux.debian.kernel


csiph-web