lkml.org 
[lkml]   [2011]   [Aug]   [26]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    From
    Date
    SubjectRe: [PATCH 1/2] x86, ioapic: Reserve only 128 bytes for IOAPICs
    On Fri, Aug 26, 2011 at 12:22 AM, Cyrill Gorcunov <gorcunov@gmail.com> wrote:
    > On Thu, Aug 25, 2011 at 06:17:06PM -0600, Bjorn Helgaas wrote:
    >> On Thu, Aug 25, 2011 at 5:33 PM, Suresh Siddha
    >> <suresh.b.siddha@intel.com> wrote:
    >> > On Thu, 2011-08-25 at 16:05 -0700, Bjorn Helgaas wrote:
    >> >> Previously we reserved 1024 bytes, but that's more space than the IOAPIC
    >> >> consumes, and it can cause conflicts with nearby devices.  The known
    >> >> requirement is 68 bytes (sizeof(struct io_apic)), and rounding up to a
    >> >> power-of-2 gives us 128.
    >> >>
    >> >
    >> > Bjorn, Given the info from Intel that most of its io-apic
    >> > implementations has registers up to 0xff offset (reserved), does
    >> > reserving just the 128 bytes for the io-apic cause any address conflicts
    >> > if the next 128 bytes are allocated (by the OS) for any other device.
    >>
    >> If the OS allocated the next 128 bytes to another device, it sounds
    >> like it would cause a conflict on Intel boxes.  This must be an area
    >> that differs between vendors.  I haven't seen a spec that mentions 256
    >> bytes as the required minimum MMIO size for IOAPICs, and apparently
    >> the AMD IOAPIC decodes 240 bytes or fewer.
    >>
    >
    > Hi Bjorn,
    >
    > the former idea (as far as I remember) of all this IO_APIC_SLOT_SIZE
    > was to be sure the io-apics are allocated with 1K step (which
    > is requirements for io-apics), but definitely it doesn't consume
    > that much space neither it decode the whole range.
    >
    > Which means, I would prefer if we have (since we change IO_APIC_SLOT_SIZE
    > anyway) some additional check and WARN_ON in this code. Something like
    >
    >  if (io-apic-base-address & 0x3ff)
    >   WARN_ON();
    >
    > Hm? (also we have bad_ioapic() check, probably should put such test
    > there instead).

    Is there some spec that requires all IOAPICs to be 1K aligned? I
    don't doubt that's the case; I'd just like to see something more
    concrete than folklore. I'm pretty sure there's some (possibly
    secret) "IOAPIC architecture spec," and a section reference to it
    would be nice. Even before my patch, I don't think we actually
    checked or enforced any *alignment* -- we only set the size. I don't
    know if it's worth it unless we have a problem it would fix, and it's
    conceivable that we'd start warning about a perfectly functional
    IOAPIC that's 128-byte aligned.

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

    \
     
     \ /
      Last update: 2011-08-26 18:23    [W:8.684 / U:0.048 seconds]
    ©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site