From: "Michael S. Tsirkin" <mst@redhat.com>
To: Igor Mammedov <imammedo@redhat.com>
Cc: pbonzini@redhat.com, kevin@koconnor.net, seabios@seabios.org,
kraxel@redhat.com, qemu-devel@nongnu.org
Subject: Re: [Qemu-devel] [PATCH v2] map 64-bit PCI BARs at location provided by emulator
Date: Sun, 13 Oct 2013 18:59:20 +0300 [thread overview]
Message-ID: <20131013155920.GC11448@redhat.com> (raw)
In-Reply-To: <20131013171154.1abb4542@thinkpad>
On Sun, Oct 13, 2013 at 05:11:54PM +0200, Igor Mammedov wrote:
> On Sun, 13 Oct 2013 15:31:23 +0300
> "Michael S. Tsirkin" <mst@redhat.com> wrote:
>
> > On Sun, Oct 13, 2013 at 02:13:44PM +0200, Igor Mammedov wrote:
> > > Currently 64-bit PCI BARs are unconditionally mapped by BIOS right
> > > over 4G + RamSizeOver4G location, which doesn't allow to reserve
> > > extra space before 64-bit PCI window. For memory hotplug an extra
> > > RAM space might be reserved after present 64-bit RAM end and BIOS
> > > should map 64-bit PCI BARs after it.
> > >
> > > Introduce "etc/pcimem64-start" romfile to provide BIOS a hint
> > > where it should start mapping of 64-bit PCI BARs. If romfile is
> > > missing, BIOS reverts to legacy behavior and starts mapping right
> > > after high memory.
> > >
> > > Signed-off-by: Igor Mammedov <imammedo@redhat.com>
> > > ---
> > > v2:
> > > * place 64-bit window behind high RAM end if "etc/pcimem64-start"
> > > points below it.
> >
> > Hmm I had an alternative suggestion of passing smbios
> > tables in from QEMU (seabios already has a mechanism for this).
> > We could then simply increase the existing RamSize variable.
> >
> > What do you think about this idea?
> There is several issues with RamSizeOver4G I see:
> 1. As Gerd pointed out it is also used to setup a part of e820 tables.
> It possibly could be also changed see "[Qemu-devel] [RfC PATCH] e820: pass
> high memory too" but it will be more complicated to keep seabios
> backward/forward compatible if we use RamSizeOver4G for passing.
What are the compatibility worries, exactly?
The way I see it, if QEMU wants to be compatible it can't
use memory hotplug. In that case RamSizeOver4G will have
the old meaning which should work without issues.
> 2. It doesn't scale if we are going to use more than 1Tb future.
> Extending RamSizeOver4G value via additional CMOS port wasn't welcomed
> either, see
> http://www.seabios.org/pipermail/seabios/2010-September/000911.html
>
> It looks like fw_cfg interface is preferred one and "etc/pcimem64-start"
> approach doesn't have backward/forward compatibility issues.
Hmm but you aren't passing in memory size - pcimem64-start
merely tells bios where to put PCI devices.
So how would it help?
> >
> >
> > > ---
> > > src/fw/pciinit.c | 13 ++++++++++++-
> > > 1 file changed, 12 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/src/fw/pciinit.c b/src/fw/pciinit.c
> > > index b29db99..798309b 100644
> > > --- a/src/fw/pciinit.c
> > > +++ b/src/fw/pciinit.c
> > > @@ -18,6 +18,8 @@
> > > #include "paravirt.h" // RamSize
> > > #include "string.h" // memset
> > > #include "util.h" // pci_setup
> > > +#include "byteorder.h" // le64_to_cpu
> > > +#include "romfile.h" // romfile_loadint
> > >
> > > #define PCI_DEVICE_MEM_MIN 0x1000
> > > #define PCI_BRIDGE_IO_MIN 0x1000
> > > @@ -764,6 +766,15 @@ static void pci_bios_map_devices(struct pci_bus *busses)
> > > {
> > > if (pci_bios_init_root_regions(busses)) {
> > > struct pci_region r64_mem, r64_pref;
> > > + u64 ram64_end = 0x100000000ULL + RamSizeOver4G;
> > > + u64 base64 = le64_to_cpu(romfile_loadint("etc/pcimem64-start",
> > > + ram64_end));
> > > + if (base64 < ram64_end) {
> > > + dprintf(1, "ignorig etc/pcimem64-start [0x%llx] below present RAM, "
> > > + "placing 64-bit PCI window behind RAM end: %llx",
> > > + base64, ram64_end);
> > > + base64 = ram64_end;
> > > + }
> > > r64_mem.list.first = NULL;
> > > r64_pref.list.first = NULL;
> > > pci_region_migrate_64bit_entries(&busses[0].r[PCI_REGION_TYPE_MEM],
> > > @@ -779,7 +790,7 @@ static void pci_bios_map_devices(struct pci_bus *busses)
> > > u64 align_mem = pci_region_align(&r64_mem);
> > > u64 align_pref = pci_region_align(&r64_pref);
> > >
> > > - r64_mem.base = ALIGN(0x100000000LL + RamSizeOver4G, align_mem);
> > > + r64_mem.base = ALIGN(base64, align_mem);
> > > r64_pref.base = ALIGN(r64_mem.base + sum_mem, align_pref);
> > > pcimem64_start = r64_mem.base;
> > > pcimem64_end = r64_pref.base + sum_pref;
> > > --
> > > 1.8.3.1
> >
>
>
> --
> Regards,
> Igor
next prev parent reply other threads:[~2013-10-13 15:56 UTC|newest]
Thread overview: 30+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-10-13 12:13 [Qemu-devel] [PATCH v2] map 64-bit PCI BARs at location provided by emulator Igor Mammedov
2013-10-13 12:31 ` Michael S. Tsirkin
2013-10-13 15:11 ` Igor Mammedov
2013-10-13 15:59 ` Michael S. Tsirkin [this message]
2013-10-13 16:23 ` Igor Mammedov
2013-10-13 16:46 ` Michael S. Tsirkin
2013-10-13 17:33 ` Igor Mammedov
2013-10-13 18:19 ` [Qemu-devel] [SeaBIOS] " Igor Mammedov
2013-10-13 19:53 ` Kevin O'Connor
2013-10-14 8:01 ` Gerd Hoffmann
2013-10-13 20:28 ` [Qemu-devel] " Michael S. Tsirkin
2013-10-14 10:27 ` Igor Mammedov
2013-10-14 11:00 ` Michael S. Tsirkin
2013-10-14 12:16 ` Gerd Hoffmann
2013-10-14 12:38 ` Michael S. Tsirkin
2013-10-14 13:04 ` Gerd Hoffmann
2013-10-14 14:00 ` Michael S. Tsirkin
2013-10-14 16:15 ` Igor Mammedov
2013-10-14 16:37 ` Michael S. Tsirkin
2013-10-15 8:01 ` Gerd Hoffmann
2013-10-15 9:05 ` Igor Mammedov
2013-10-15 9:14 ` [Qemu-devel] [SeaBIOS] " Gerd Hoffmann
2013-10-15 12:36 ` Igor Mammedov
2013-10-15 9:16 ` [Qemu-devel] " Michael S. Tsirkin
2013-10-15 9:24 ` Gerd Hoffmann
2013-10-15 9:53 ` Igor Mammedov
2013-10-15 9:47 ` Igor Mammedov
2013-10-15 9:08 ` Michael S. Tsirkin
2013-10-14 12:28 ` Igor Mammedov
2013-10-13 15:15 ` Kevin O'Connor
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20131013155920.GC11448@redhat.com \
--to=mst@redhat.com \
--cc=imammedo@redhat.com \
--cc=kevin@koconnor.net \
--cc=kraxel@redhat.com \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=seabios@seabios.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).