Skip to contentSkip to conversation

input completely dead in X

BUG: 2.6.9-rc2-bk11 Micha Feigin 5 participants 10 messages Sep 26, 2004 – Sep 29, 2004

lore
Share & tools
lore
#0 lore raw

Just tried kernel 2.6.9-rc2-bk11 and when I start X input is completely dead (including num-lock, caps-lock, sysrq and mouse). The computer is otherwise functional (I can log in with ssh, kill X and everything is functional again).

In the console is completely functional, before and after starting X.

I had exactly the same behavior with mm kernels from version 2.6.7 or 2.6.8 (not sure) onwards, which was fixed by reversing the bk-input patch.

kernel 2.6.9-rc2 works fine.

The machine is Sony Vaio PCG-FXA53 laptop (amd 1500+ and via chipset). The keyboard is recognized as ps2 AFAIK and the touchpad is an alps (with the alps kernel patch).

#1 lore raw

Salut,

Quoted 4 lines · Micha Feigin

Just tried kernel 2.6.9-rc2-bk11 and when I start X input is completely dead (including num-lock, caps-lock, sysrq and mouse). The computer is otherwise functional (I can log in with ssh, kill X and everything is functional again).

Which X do you use? And which version?

Tonnerre

#2 lore raw
Quoted 10 lines · Tonnerre

Salut,

Quoted 4 lines · Micha Feigin

Just tried kernel 2.6.9-rc2-bk11 and when I start X input is completely dead (including num-lock, caps-lock, sysrq and mouse). The computer is otherwise functional (I can log in with ssh, kill X and everything is functional again).

Which X do you use? And which version?

xfree 4.3 in debian unstable (I believe its somewhat modified), debian calls it 4.3.0.dfsg.1-7.

The bk-input patch is a bit big for me to search for the problematic change, but anyone has any leads I will happily poke around further.

Quoted 1 line

Tonnerre

#3 lore raw
Quoted 9 lines · Micha Feigin
Quoted 8 lines · Tonnerre

Salut,

Quoted 4 lines · Micha Feigin

Just tried kernel 2.6.9-rc2-bk11 and when I start X input is completely dead (including num-lock, caps-lock, sysrq and mouse). The computer is otherwise functional (I can log in with ssh, kill X and everything is functional again).

Quoted 9 lines
Quoted 2 lines

Which X do you use? And which version?

xfree 4.3 in debian unstable (I believe its somewhat modified), debian calls it 4.3.0.dfsg.1-7.

The bk-input patch is a bit big for me to search for the problematic change, but anyone has any leads I will happily poke around further.

Did some more testing and apparently its something to do with the alps touchpad patch. When using the alps as a touchpad all the input completely dies on X startup, when I use it as a GlidePointPS/2 things work fine (even with the alps patch), although mouse is a bit slow.

Will try and dig further and see if I come up with a fix.

Quoted 13 lines
Quoted 1 line

Tonnerre

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

+++++++++++++++++++++++++++++++++++++++++++
This Mail Was Scanned By Mail-seCure System
at the Tel-Aviv University CC.

#4 lore raw

Resending as I forgot to CC Vojtech and Peter first time around...

Quoted 6 lines · Micha Feigin
Quoted 3 lines

Or better yet, use the auto-dev feature, which should work if you have a new enough X driver and kernel patch.

auto-dev doesn't work for me and I don't have time to check it out.

Addition of Kensington ThinkingMouse / ExpertMouse support caused Synaptics and ALPS protocol numbers to move to 8 and 9 respectively which broke Peter's auto-dev detection.

Vojtech, we need to keep protcol numbers stable, I propose something like this:

enum psmouse_type {
        PSMOUSE_PS2             = 0,
        PSMOUSE_PS2PP,
        PSMOUSE_THINKPS,
        PSMOUSE_GENPS           = 64,   /* 4 byte protocol start */ PSMOUSE_IMPS,
        PSMOUSE_IMEX,
        PSMOUSE_SYNAPTICS       = 128,  /* 5+ byte protocols start */ PSMOUSE_ALPS,
};

Peter, if we adopt the scheme above you will have to check both for old and new protocol numbers; in addition you need to BTN_TOOL_FINGER device bit to make sure you are dealing with a touchpad.

Any holes here?

Signature · 1 line
Dmitry
#5 lore raw
Quoted 7 lines · Dmitry Torokhov

Resending as I forgot to CC Vojtech and Peter first time around...

Quoted 4 lines · Micha Feigin
Quoted 3 lines

Or better yet, use the auto-dev feature, which should work if you have a new enough X driver and kernel patch.

Quoted 6 lines
Quoted 2 lines

auto-dev doesn't work for me and I don't have time to check it out.

Addition of Kensington ThinkingMouse / ExpertMouse support caused Synaptics and ALPS protocol numbers to move to 8 and 9 respectively which broke Peter's auto-dev detection.

Ouch. I suspected something bad will happen.

Quoted 12 lines

Vojtech, we need to keep protcol numbers stable, I propose something like this:

enum psmouse_type {
        PSMOUSE_PS2             = 0,
        PSMOUSE_PS2PP,
        PSMOUSE_THINKPS,
        PSMOUSE_GENPS           = 64,   /* 4 byte protocol start */ PSMOUSE_IMPS,
        PSMOUSE_IMEX,
        PSMOUSE_SYNAPTICS       = 128,  /* 5+ byte protocols start */ PSMOUSE_ALPS,
};

No, we really need to keep backwards compatibility with the numbering here and solve the packetsize issue elsewhere. Probably the best would be for each of the protocols to have its own packet collection routine, like the Synaptics and ALPS already have. It could be shared among the simpler protocols.

We'll need this anyway for a heuristic resynchronizer.

Quoted 5 lines

Peter, if we adopt the scheme above you will have to check both for old and new protocol numbers; in addition you need to BTN_TOOL_FINGER device bit to make sure you are dealing with a touchpad.

Any holes here?

I really would prefer if old installations of the X driver would work with new kernel without the need to upgrade the X driver. I propose to keep the protocol numbers intact, and if possible to also find a better way for the X driver to detect a touchpad than relying on the IDs of the input device, namely based on BTN_TOOL_FINGER presence.

Signature · 2 lines
Vojtech Pavlik
SuSE Labs, SuSE CR
#6 lore raw
Quoted 30 lines · Vojtech Pavlik
Quoted 7 lines · Dmitry Torokhov

Resending as I forgot to CC Vojtech and Peter first time around...

Quoted 4 lines · Micha Feigin
Quoted 3 lines

Or better yet, use the auto-dev feature, which should work if you have a new enough X driver and kernel patch.

Quoted 6 lines
Quoted 2 lines

auto-dev doesn't work for me and I don't have time to check it out.

Addition of Kensington ThinkingMouse / ExpertMouse support caused Synaptics and ALPS protocol numbers to move to 8 and 9 respectively which broke Peter's auto-dev detection.

Ouch. I suspected something bad will happen.

Quoted 12 lines

Vojtech, we need to keep protcol numbers stable, I propose something like this:

enum psmouse_type {
        PSMOUSE_PS2             = 0,
        PSMOUSE_PS2PP,
        PSMOUSE_THINKPS,
        PSMOUSE_GENPS           = 64,   /* 4 byte protocol start */ PSMOUSE_IMPS,
        PSMOUSE_IMEX,
        PSMOUSE_SYNAPTICS       = 128,  /* 5+ byte protocols start */ PSMOUSE_ALPS,
};

Quoted 8 lines

No, we really need to keep backwards compatibility with the numbering here and solve the packetsize issue elsewhere. Probably the best would be for each of the protocols to have its own packet collection routine, like the Synaptics and ALPS already have. It could be shared among the simpler protocols.

We'll need this anyway for a heuristic resynchronizer.

Ok, for now I am killing PS2TPP which is not really useful and will allow THINKPS take it's spot moving SYNAPTICS and ALPS to their former numbers.

Signature · 1 line
Dmitry
#7 lore raw
Quoted 2 lines · Dmitry Torokhov

Ok, for now I am killing PS2TPP which is not really useful and will allow THINKPS take it's spot moving SYNAPTICS and ALPS to their former numbers.

Where's the patch? ;)

Signature · 2 lines
Vojtech Pavlik
SuSE Labs, SuSE CR
#8 lore raw
Quoted 7 lines · Vojtech Pavlik
Quoted 2 lines · Dmitry Torokhov

Ok, for now I am killing PS2TPP which is not really useful and will allow THINKPS take it's spot moving SYNAPTICS and ALPS to their former numbers.

Where's the patch? ;)

Right here in my queue ;) I should have the next set of patches ready either tonight or tomorrow night.

Signature · 1 line
Dmitry
#9 lore raw
In reply to Micha Feigin
Quoted 5 lines · Micha Feigin

xfree 4.3 in debian unstable (I believe its somewhat modified), debian calls it 4.3.0.dfsg.1-7.

The bk-input patch is a bit big for me to search for the problematic change, but anyone has any leads I will happily poke around further.

Make sure that the Debian developers have included the patches/configuration options so that the keyboard driver is not banging the keyboard/mouse I/O ports directly. The input layer does not like that, although I've yet to see a crash/hang from it.

Alan