Messages in this thread | | | From | Bjorn Helgaas <> | Subject | Re: [PATCH] x86: tidy e820 output | Date | Wed, 22 Sep 2010 12:20:33 -0600 |
| |
On Wednesday, September 22, 2010 11:54:49 am H. Peter Anvin wrote: > On 09/22/2010 10:27 AM, Bjorn Helgaas wrote: > > > > This tidies e820 output by adding an "e820" prefix and printing ranges in > > the same style we use for struct resource with %pR, e.g.: > > > > - BIOS-e820: 0000000000000000 - 000000000009f400 (usable) > > + BIOS-e820: [mem 0x00000000-0x0009f3ff] (usable) > > I'm sorry, I have to admit to not understanding the difference. I do > not want to change the number of hex digits from fixed 16 digits, as > that will make the output harder to read when printed in a block (as is > normal for the early e820 dump). The [mem] prefix seems redundant with > (usable), or am I misreading this?
These E820 ranges should be easily comparable with similar ranges we print elsewhere. Currently we have things like this:
BIOS-e820: 0000000000000000 - 000000000009f400 (usable) BIOS-e820: 000000000009f400 - 00000000000a0000 (reserved) BIOS-e820: 00000000fffbc000 - 0000000100000000 (reserved) pci_root PNP0A03:00: host bridge window [mem 0x000a0000-0x000bffff] pci_root PNP0A03:00: host bridge window [mem 0xe0000000-0xfebfffff] pci 0000:00:02.0: reg 10: [mem 0xf0000000-0xf1ffffff pref] reserve RAM buffer: 000000000009f400 - 000000000009ffff pnp 00:07: [mem 0xfed00000-0xfed003ff]
It would be easier to integrate the E820 information with the ACPI and PCI window and BAR information if they looked similar.
We currently have a mix of some with "0x" prefix, some without; some with eight hex digits, some with sixteen; some with spaces around the internal "-", some without; some with type (io/mem/etc), some without; some with uppercase hex (MTRR), most with lowercase; some including the end address, some not; and even some in PFNs and most in addresses. It just makes it harder than it needs to be to debug issues in this area.
Bjorn
| |