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

Re: [PATCH v2] xen/efi: Handle efi_memmap with Xen relocating itself


  • To: Jan Beulich <jbeulich@xxxxxxxx>
  • From: Jason Andryuk <jason.andryuk@xxxxxxx>
  • Date: Wed, 18 Nov 2026 19:22:27 -0500
  • Arc-authentication-results: i=1; mx.microsoft.com 1; spf=pass (sender ip is 165.204.84.17) smtp.rcpttodomain=suse.com smtp.mailfrom=amd.com; dmarc=pass (p=quarantine sp=quarantine pct=100) action=none header.from=amd.com; dkim=none (message not signed); arc=none (0)
  • Arc-message-signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=microsoft.com; s=arcselector10001; h=From:Date:Subject:Message-ID:Content-Type:MIME-Version:X-MS-Exchange-AntiSpam-MessageData-ChunkCount:X-MS-Exchange-AntiSpam-MessageData-0:X-MS-Exchange-AntiSpam-MessageData-1; bh=K60405PsVnpFO5oZvPCRUFrgBeQMbHQfmkyDAIKnsRo=; b=eaHOX6yKTOytEGnjt19IP7tfNMjgbhQTaTHdGJLnuHZaUgaxo/rjEFOh/kWyf+JaQAuLF11MxFP+cGsbTPpNl7ORTQM+m8XvhKrR0GoFFmiCqzAcWxkGqWi6n+OXSeqKhIZL2xnJ0JkvsaVNds6vwaZQHvAkXItgHQtYmrNiSFkmrWAlsgmQHq4ksaJbBxYP+gq0QldpXQR2NCffSwQRgyTfLRWKiuc5ntlQm/HezeK+/sAtcNaj40hirNoZDkRYWNJR6vmWlz4DTcqyhr5+dYlfMe1ngze5tY5H+z6quDIHSRG9xpZQfOsRKvWAXFLh907tL1zO075nlqJ6PMAzFA==
  • Arc-seal: i=1; a=rsa-sha256; s=arcselector10001; d=microsoft.com; cv=none; b=o8ixKNk5J2CSYCxtEXlak3MZqudcbNtm+ohDsWqW9lCne4HBzB45FLfn2tj+KbzF6z+3NExQ12jGkEnC3hnG2bOyUkRs6zPr2HFjKFHh+XKU/az0FHtFNe63JKuE57UioZBLZY0FwS445iOKJrjh5WI1vHx+OkS0CCY3kkpCSkrc81gk+ClbmhiKCHhdkHwKvWuofEdPO/e2yZjjhIpA7fhFhV6p4ZaZAe0FxkSgsmJI5H/u2e1rGkGJlCs5CCIsQ1hZy6gil5l4Q5otxw3m5dsAQdIFJGwt2vs+lkrvuUUEwnuxQar0LqnZcNUcmcOTzVXlFPTXP8euHl+yNMGnEg==
  • Authentication-results: eu.smtp.expurgate.cloud; dkim=pass header.s=selector1 header.d=amd.com header.i="@amd.com" header.h="From:Date:Subject:Message-ID:Content-Type:MIME-Version:X-MS-Exchange-SenderADCheck"
  • Cc: Stefano Stabellini <sstabellini@xxxxxxxxxx>, Julien Grall <julien@xxxxxxx>, Bertrand Marquis <bertrand.marquis@xxxxxxx>, Michal Orzel <michal.orzel@xxxxxxx>, Volodymyr Babchuk <Volodymyr_Babchuk@xxxxxxxx>, "Daniel P. Smith" <dpsmith@xxxxxxxxxxxxxxxxxxxx>, Marek Marczykowski-Górecki <marmarek@xxxxxxxxxxxxxxxxxxxxxx>, Andrew Cooper <andrew.cooper3@xxxxxxxxxx>, Roger Pau Monné <roger@xxxxxxxxxxxxxx>, Teddy Astie <teddy.astie@xxxxxxxxxx>, <xen-devel@xxxxxxxxxxxxxxxxxxxx>
  • Delivery-date: Thu, 19 Nov 2026 00:23:04 +0000
  • List-id: Xen developer discussion <xen-devel.lists.xenproject.org>

On 2026-11-18 03:14, Jan Beulich wrote:
On 18.11.2026 01:50, Jason Andryuk wrote:
With Xen booting via Multiboot2/GRUB on an EFI platform, and modified to
use MULTIBOOT2_LOAD_PREFERENCE_LOW, Xen relocates itself with
move_xen().

However a PVH dom0 shows:
kernel: Failed to lookup header 0x77ffd000 in Xen memory map: error -3
kernel: ACPI=0x77ffd000

... for all the Xen efi memory map lookups.  Dumping the memory map at
runtime showed it was scrubbed.  Because Xen relocated itself, the
original physical address referenced through a directmap address goes
stale.

Add efi_arch_mmap_buffer_virt() to convert from the initial value to an
appropriate runtime value.  ARM uses Boot Services AllocatePool(), so it
must continue to use DIRECTMAP_VIRT_START.

For x86, we use an offset into the loaded Xen binary at the time of the
call and add the to XEN_VIRT_START.  ebmalloc() allocates from the Xen
image memory, so it moves during relocation.  xen_phys_start is set for
xen.efi in efi_exit_boot(), but not for xen.gz, so it cannot be used.

Add a generic helper using `lea __image_base__(%rip)` to get the current
load address, and use that to calculate the offset into the Xen image.
That offset is also valid from XEN_VIRT_START.  Ensure the value passed
in is within the image.  Use the generic helper to implement
efi_arch_mmap_buffer_virt().

Fixes: c5b9805bc1f7 ("efi: create new early memory allocator")
Signed-off-by: Jason Andryuk <jason.andryuk@xxxxxxx>
---
I think it could be seen without MULTIBOOT2_LOAD_PREFERENCE_LOW if xen
chose to relocate itself.  But since grub usually loads xen high, it
doesn't relocate itself.

_end is already calculated as a rip-relative address by the compiler.  I
still added an inline asm to ensure a rip-relative value is used.  i.e.
I don't see why the compiler couldn't insert the absolute value, if it
chose to.

It can't, as then it would violate the -fpic that we pass to it. IOW it's
really the other way around: In asm()-s we need to be careful not to
violate compiler options in effect.

Right.  Thank you.

--- a/xen/arch/x86/efi/efi-boot.h
+++ b/xen/arch/x86/efi/efi-boot.h
@@ -219,11 +219,35 @@ static void __init 
efi_arch_process_memory_map(EFI_SYSTEM_TABLE *SystemTable,
} +/*
+ * Convert a runtime 1:1 physical address within the
+ * Xen image to a Xen virtual address.
+ */
+static void *__init efi_arch_image_phys_to_virt(void *ptr)
+{
+    unsigned long load_base;
+    unsigned long load_end;
+
+    /* RIP-relative address is the current Xen load address. */
+    asm ( "lea __image_base__(%%rip), %0" : "=r" (load_base) );
+    asm ( "lea _end(%%rip), %0" : "=r" (load_end) );
+
+    ASSERT(ptr >= (void *)load_base);

Hmm, is __image_base__ really the right anchor? move_xen()'s memcpy()
only covers [_start,_end).
Yes, you are correct and _start is a better limit.

With your comment above about -fpic, _start and _end will be RIP-relative. We can avoid the inline asm and do:

static void *__init efi_arch_image_phys_to_virt(void *ptr)
{
    unsigned long phys_start = (unsigned long)_start;
    unsigned long virt_start = XEN_VIRT_START + XEN_IMG_OFFSET;

    ASSERT(ptr >= (void *)_start);
    ASSERT(ptr <  (void *)_end);

    return ptr - phys_start + virt_start;
}

Avoiding the inline asm is nice, but retaining the inline asm make it explicit. Thoughts?

Regards,
Jason



 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.