Comment 17 for bug 311254

> diff --git a/dix/getevents.c b/dix/getevents.c
> index 707d1da..16e23dc 100644
> --- a/dix/getevents.c
> +++ b/dix/getevents.c
> @@ -695,6 +695,9 @@ positionSprite(DeviceIntPtr dev, int *x, int *y,
> dev->valuator->axes + 1, scr->height);
> }
>
> + /* dropy x/y (device coordinates) back into valuators for next event */
> + dev->last.valuators[0] = *x;
> + dev->last.valuators[1] = *y;
> }
>
> /**
> @@ -980,9 +983,6 @@ GetPointerEvents(EventList *events, DeviceIntPtr pDev, int
> type, int buttons,
> positionSprite(pDev, &x, &y, scr, &cx, &cy);
> updateHistory(pDev, first_valuator, num_valuators, ms);
>
> - /* dropy x/y (device coordinates) back into valuators for next event */
> - pDev->last.valuators[0] = x;
> - pDev->last.valuators[1] = y;
>
> /* Update the valuators with the true value sent to the client*/
> if (num_valuators >= 1 && first_valuator == 0)

I really fail to see how that patch has to do with the crash you described,
especially since GPE isn't touched at all on key events.

> Backtrace:
> 0: X(xorg_backtrace+0x3b) [0x812dbbb]
> 1: X(xf86SigHandler+0x51) [0x80c8bc1]
> 2: [0xffffe400]
> 3: X(mieqProcessInputEvents+0x327) [0x810d1e7]
> 4: X(ProcessInputEvents+0x17) [0x80c9787]
> 5: X(Dispatch+0x6e) [0x808716e]
> 6: X(main+0x3bd) [0x806c47d]
> 7: /lib/i686/libc.so.6(__libc_start_main+0xe5) [0xb7c0e5c5]
> 8: X [0x806b931]

this backtrace is unrelated to the patch above as well.