Created attachment 12054 [details] [review]
dummy functions

With the attached patch, the backtrace looks like this:

(gdb) bt full
#0  0x0818f9b8 in RRCrtcSetRotations (crtc=0x0, rotations=1) at rrcrtc.c:108
No locals.
#1  0x0810d364 in xf86RandR12SetRotations (pScreen=0x826de00, rotations=1)
    at xf86RandR12.c:535
        crtc = 0x826d800
        randrp = 0x0
        pScrn = 0x826f000
        c = 0
        config = 0x8263480
#2  0x08105d24 in xf86CrtcScreenInit (screen=0x826de00) at xf86Crtc.c:684
        scrn = 0x826f000
        config = 0x8263480
        c = 0
#3  0x286e2ea9 in MGAScreenInit (scrnIndex=0, pScreen=0x826de00, argc=1, 
    argv=0xbfbfec0c) at mga_driver.c:2907
        pScrn = 0x826f000
        hwp = 0x8271000
        pMga = 0x8270000
        MGAdac = 0x82700d8
        ret = 1
        visual = 0x8278fdc
        FBStart = (unsigned char *) 0x287e1000 " \a"
        width = 1280
---Type <return> to continue, or q <return> to quit--- 
        height = 1024
        displayWidth = 1280
        f = 3
        VRTemp = 217579520
        FBTemp = 136272172
        driFrom = X_PROBED
#4  0x0806e775 in AddScreen (pfnInit=0x286e23d8 <MGAScreenInit>, argc=1, 
    argv=0xbfbfec0c) at main.c:769
        i = 0
        scanlinepad = 32
        format = 7
        depth = 32
        bitsPerPixel = 32
        j = 2
        k = 2
        pScreen = 0x826de00
#5  0x080ae75b in InitOutput (pScreenInfo=0x8247e80, argc=1, argv=0xbfbfec0c)
    at xf86Init.c:840
        i = 0
        j = 0
        k = 673303337
        scr_index = -1077941320
        generation = 1
---Type <return> to continue, or q <return> to quit---
        modulelist = (char **) 0x8267390
        optionlist = (pointer *) 0x8255600
        layout = 0x824c738
        screenpix24 = Pix24DontCare
        pix24 = Pix24Use32
        pix24From = X_DEFAULT
        pix24Fail = 0
        autoconfig = 0
#6  0x0806d8eb in main (argc=1, argv=0xbfbfec0c, envp=0xbfbfec14) at main.c:369
        i = 256
        j = 2
        k = 2
        error = -1077941228
        xauthfile = 0x0
        alwaysCheckForInput = {0, 1}


Most likely one of the dummies should not be a dummy, but this is now over my head.

I pushed these changes that we worked out two weeks ago (with the proper return value of crtc_lock(). Next week I really should have time to try out the card myself and fix the remaining bugs. That backtrace looks like something less trivial is wrong though.

Mass closure: This bug has been untouched for more than six years, and is not
obviously still valid. Please reopen this bug or file a new report if you continue to experience issues with current releases.