Re: [Qemu-devel] [FYI] Soft feature freeze for 1.0 is 10/15 (three weeks away)

[Alexander Graf]

On 27.09.2011, at 21:19, Blue Swirl wrote:

On Tue, Sep 27, 2011 at 4:44 PM, Avi Kivity <address@hidden> wrote:

On 09/27/2011 07:39 PM, Blue Swirl wrote:

 Well, it's not that easy.  As the other mapping is part of an ordinary

BAR,

 you need to setup the device (at least PCI_COMMAND and

PCI_BASE_ADDRESS_0)

 so it responds to memory requests, and also enable the bridge.

 We could hack it by having a low-priority mapping at 0x80013000, but it

 seems wrong.  Maybe the firmware should configure that BAR first?  What

 happens on real hardware?

In this message I seem to confess that the address is arbitrary and in

the subsequent messages the overlap with PCI region is also discussed.

http://lists.nongnu.org/archive/html/qemu-devel/2009-01/msg00542.html

Maybe the address of macio should be fixed as Laurent suggested.

I'll leave it up to you - I'm out of my depth here.

Meanwhile I suggest applying the pci alias patch - the bug is independent of

the vga issue.

OK, pushed.

Ah, after digging into the issue for a while myself I stumbled over this thread, explaining what I figured would be the culprit. Please CC qemu-ppc for these discussions :).

Either way, I'm seeing 2 issues here. I enhanced the "info mtree" command to also display the region we're currently looking at, so the first tuple shows the region we're in, the second tuple the region we're trying to map into the first region (addresses aligned on each other). I also print out alias information directly:

[ 00000000-7ffffffffffffffe ] 00000000-7ffffffffffffffe : system

  [ 80013000-8001303f ] 80013000-8001303f : escc

  [ fee00000-fee00fff ] fee00000-fee00fff : pci-data-idx

  [ fec00000-fec00fff ] fec00000-fec00fff : pci-conf-idx

  [ 80000000-fdffffff ] 80000000-fdffffff : pci-hole

   -> alias (0000000080000000, 0000000000000000)

    [ 80000000-fdffffff ] 00000000-ffffffff : pci-mmio

      [ 80880000-808fffff ] 80880000-808fffff : macio

        [ 808e0000-808fffff ] 808e0000-808fffff : macio-nvram

        [ 808a0000-808a0fff ] 808a0000-808a0fff : pmac-ide

        [ 80896000-80897fff ] 80896000-80897fff : cuda

        [ 80893000-8089303f ] 80893000-8089303f : escc-bar

         -> alias (0000000000000000, 0000000080013000)

          [ 80893000-8089303f ] 80893000-8089303f : escc

        [ 80888000-80888fff ] 80888000-80888fff : dbdma

        [ 80880000-80880fff ] 80880000-80880fff : heathrow-pic

      [ 80800000-8080ffff ] 80800000-8080ffff : vga.rom

      [ 80000000-807fffff ] 80000000-807fffff : vga.vram

  [ fe000000-fe1fffff ] fe000000-fe1fffff : isa-mmio

I/O

[ 00000000-0000ffff ] 00000000-0000ffff : io

  [ 00000700-0000070f ] 00000700-0000070f : cmd646-bmdma

    [ 0000070c-0000070f ] 0000070c-0000070f : cmd646-bmdma-ioport

    [ 00000708-0000070b ] 00000708-0000070b : cmd646-bmdma-bus

    [ 00000704-00000707 ] 00000704-00000707 : cmd646-bmdma-ioport

    [ 00000700-00000703 ] 00000700-00000703 : cmd646-bmdma-bus

  [ 00000680-00000683 ] 00000680-00000683 : cmd646-cmd

  [ 00000600-00000607 ] 00000600-00000607 : cmd646-data

  [ 00000580-00000583 ] 00000580-00000583 : cmd646-cmd

  [ 00000500-00000507 ] 00000500-00000507 : cmd646-data

  [ 00000400-000004ff ] 00000400-000004ff : ne2000

I did another small hack to display the flat memory view:

  ranges[0] = { 80000000, 13000 } vga.vram

  ranges[1] = { 80013000, 40 } escc

  ranges[2] = { 80013040, 7ecfc0 } vga.vram

  ranges[3] = { 80800000, 10000 } vga.rom

  ranges[4] = { 80880000, 1000 } heathrow-pic

  ranges[5] = { 80888000, 1000 } dbdma

  ranges[6] = { 80893000, 40 } escc

  ranges[7] = { 80896000, 2000 } cuda

  ranges[8] = { 808a0000, 1000 } pmac-ide

  ranges[9] = { 808e0000, 20000 } macio-nvram

  ranges[10] = { fe000000, 200000 } isa-mmio

  ranges[11] = { fec00000, 1000 } pci-conf-idx

  ranges[12] = { fee00000, 1000 } pci-data-idx

Now to the issues:

1)

ESCC is mapped inside the VGA region. That's what you discussed in this thread. It's wrong. Please check this bug for a dump of a real G3 Beige's memory layout:

  http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=555651

I don't think we really need to have serial available before PCI enum, so let's just ditch the first map.

2)

vga.vram continues to be mapped, but apparently isn't accessible. I would expect the hole of 40 bytes to be non-accessible / broken, but what ends up happening is that the whole second region apparently is unusable. What exactly is going on there? Sounds like a memory API bug to me.

Alex