Re: Mapping Video BIOS?

From: Greg 'groggy' Lehey (grog_at_FreeBSD.org)
Date: 07/27/03

  • Next message: Dan Pelleg: "Re: MD5 signatures of 5.1-RELEASE" 
    Date: Sun, 27 Jul 2003 10:39:38 +0930
To: "M. Warner Losh" <imp@bsdimp.com>

    On Saturday, 26 July 2003 at 18:44:43 -0600, M. Warner Losh wrote:
    > In message: <20030727002138.GD45069@wantadilla.lemis.com>
    > "Greg 'groggy' Lehey" <grog@FreeBSD.org> writes:
    >> On Saturday, 26 July 2003 at 11:27:06 -0600, M. Warner Losh wrote:
    >>> In message: <20030726080217.GB45069@wantadilla.lemis.com>
    >>> "Greg 'groggy' Lehey" <grog@freebsd.org> writes:
    >>>> machine doesn't have a serial port, so I can't apply a kernel debugger
    >>>> to find out what's going on.
    >>>
    >>> Does it have a firewire port?
    >>
    >> Yes. How can I use that?
    >
    > If you have a second machine with firewire, then you can use the
    > firewire port as your console. Look at /usr/ports/devel/dcons. It is
    > one of the under-publicized cool features from Japan (Thanks
    > Shimokawa-san!).

    Ah, good stuff. I'll have to check if it also works with gdb.
    Unfortunately, this is my only machine with firewire. I was wondering
    if there were USB/conventional serial converters that I could use.

    >> I had also expected that you could shed some light on the BIOS mapping
    >> issue. Since my last message I've become pretty sure that it must be
    >> something to do with the chip set setup. Is it possible that we're
    >> not mapping the entire area 0xc0000 to 0xfffff?
    >
    > I'm not sure what you mean by this question. Since OLDCARD works, and
    > requires read/write access to that physical memory range, I doubt that
    > it is unmapped.

    I'm not sure at what level. I suspect that something in the chipset
    is turning off that area of memory, or mapping something else to it.
    The dump from Microsoft shows that there's another BIOS at 0xcf000,
    but what I have mapped in memory shows only 0xff up to address
    0xd0000, where I find another BIOS signature:

    0x28377fe0: 0xffffffff 0xffffffff 0xffffffff 0xffffffff
    0x28377ff0: 0xffffffff 0xffffffff 0xffffffff 0xffffffff
    0x28378000: 0xe80caa55 0x4ecb14c8 0x0000033b 0x00000000
    0x28378010: 0x00000000 0x00200000 0x00600040 0x90c08b2e
    0x28378020: 0x49444e55 0x0000ea16 0x0c9d0201 0xad100800

    > It may be the case that we aren't setting things up so that XFree86
    > can call the BIOS, but given that we used PCIBIOS before ACPI, it
    > seems unlikely.

    Well, this is a new laptop, so it's possible that something *is*
    getting set up incorrectly.

    Greg 

    --
See complete headers for address and phone numbers

    • application/pgp-signature attachment: stored
  • Next message: Dan Pelleg: "Re: MD5 signatures of 5.1-RELEASE"

    Relevant Pages

    • Re: Mapping Video BIOS?
      ... machine doesn't have a serial port, so I can't apply a kernel debugger ... :> If you have a second machine with firewire, ... :>> I had also expected that you could shed some light on the BIOS mapping ... is turning off that area of memory, or mapping something else to it. ...
      (freebsd-current)
    • Re: [Full-disclosure] Firewire Attack on Windows Vista
      ... Or, if one is already using a firewire device, but has ... "freeze" the memory chip, take it out, put it in their laptop, search ... Firewire chipsets allow drivers to configure a particular memory range ... which is open to access by DMA devices. ...
      (Full-Disclosure)
    • RE: [Full-disclosure] Firewire Attack on Windows Vista
      ... Or, if one is already using a firewire device, but has ... "freeze" the memory chip, take it out, put it in their laptop, search ... Firewire chipsets allow drivers to configure a particular memory range ... which is open to access by DMA devices. ...
      (Bugtraq)
    • Re: Physical Memory Snapshot
      ... If I use this PCI device and say I want to ... Can I be sure that the DMA is atomic? ... the device uses the memory bus, there won't be any writes from the CPU ... This can be avoided by dumping the memory remotely via FireWire. ...
      (Linux-Kernel)
    • FireWire Security issues
      ... memory, bypassing operating system limitations. ... physical security measures if they are equipped with "FireWire" ports. ... Systems that counted on physical access limitation such as blocking access ... involvement of the operating system on the host computer. ...
      (FreeBSD-Security)