qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH v2 0/4] hw/arm/virt: Improve address assignment for high memo


From: Gavin Shan
Subject: Re: [PATCH v2 0/4] hw/arm/virt: Improve address assignment for high memory regions
Date: Fri, 26 Aug 2022 16:02:47 +1000
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.2.0

Hi Eric,

On 8/24/22 6:06 PM, Eric Auger wrote:
On 8/24/22 05:29, Gavin Shan wrote:
On 8/15/22 4:29 PM, Gavin Shan wrote:
There are three high memory regions, which are VIRT_HIGH_REDIST2,
VIRT_HIGH_PCIE_ECAM and VIRT_HIGH_PCIE_MMIO. Their base addresses
are floating on highest RAM address. However, they can be disabled
in several cases.
      (1) One specific high memory region is disabled by developer by
      toggling vms->highmem_{redists, ecam, mmio}.
      (2) VIRT_HIGH_PCIE_ECAM region is disabled on machine, which is
      'virt-2.12' or ealier than it.
      (3) VIRT_HIGH_PCIE_ECAM region is disabled when firmware is loaded
      on 32-bits system.
      (4) One specific high memory region is disabled when it breaks the
      PA space limit.
      The current implementation of virt_set_memmap() isn't comprehensive
because the space for one specific high memory region is always
reserved from the PA space for case (1), (2) and (3). In the code,
'base' and 'vms->highest_gpa' are always increased for those three
cases. It's unnecessary since the assigned space of the disabled
high memory region won't be used afterwards.

The series intends to improve the address assignment for these
high memory regions:

PATCH[1] and PATCH[2] are cleanup and preparatory works.
PATCH[3] improves address assignment for these high memory regions
PATCH[4] moves the address assignment logic into standalone helper

History
=======
v1: https://lists.nongnu.org/archive/html/qemu-arm/2022-08/msg00013.html

Changelog
=========
v2:
    * Split the patches for easier review                        (Gavin)
    * Improved changelog                                         (Marc)
    * Use 'bool fits' in virt_set_high_memmap()                  (Eric)

You did not really convince me about migration compat wrt the high MMIO
region. Aren't the PCI BARs saved/restored meaning the device driver is
expecting to find data at the same GPA. But what if your high MMIO
region was relocated in the dest QEMU with a possibly smaller VM IPA?
Don't you have MMIO regions now allocated outside of the dest MMIO
region? How does the PCI host bridge route accesses to those regions?
What do I miss?


I'm currently looking into virtio-pci-net migration, but need time to
investigate how the device is migrated. I will get back to you once
I have something. Thanks for your comments :)

Thanks,
Gavin




reply via email to

[Prev in Thread] Current Thread [Next in Thread]