releases/5.0.19/fbdev-efifb-ignore-framebuffer-memmap-entries-that-lack-any-memory-types.patch

   1 From f8585539df0a1527c78b5d760665c89fe1c105a9 Mon Sep 17 00:00:00 2001
   2 From: Ard Biesheuvel <ard.biesheuvel@linaro.org>
   3 Date: Thu, 16 May 2019 23:31:59 +0200
   4 Subject: fbdev/efifb: Ignore framebuffer memmap entries that lack any memory types
   5
   6 From: Ard Biesheuvel <ard.biesheuvel@linaro.org>
   7
   8 commit f8585539df0a1527c78b5d760665c89fe1c105a9 upstream.
   9
  10 The following commit:
  11
  12   38ac0287b7f4 ("fbdev/efifb: Honour UEFI memory map attributes when mapping the FB")
  13
  14 updated the EFI framebuffer code to use memory mappings for the linear
  15 framebuffer that are permitted by the memory attributes described by the
  16 EFI memory map for the particular region, if the framebuffer happens to
  17 be covered by the EFI memory map (which is typically only the case for
  18 framebuffers in shared memory). This is required since non-x86 systems
  19 may require cacheable attributes for memory mappings that are shared
  20 with other masters (such as GPUs), and this information cannot be
  21 described by the Graphics Output Protocol (GOP) EFI protocol itself,
  22 and so we rely on the EFI memory map for this.
  23
  24 As reported by James, this breaks some x86 systems:
  25
  26   [ 1.173368] efifb: probing for efifb
  27   [ 1.173386] efifb: abort, cannot remap video memory 0x1d5000 @ 0xcf800000
  28   [ 1.173395] Trying to free nonexistent resource <00000000cf800000-00000000cf9d4bff>
  29   [ 1.173413] efi-framebuffer: probe of efi-framebuffer.0 failed with error -5
  30
  31 The problem turns out to be that the memory map entry that describes the
  32 framebuffer has no memory attributes listed at all, and so we end up with
  33 a mem_flags value of 0x0.
  34
  35 So work around this by ensuring that the memory map entry's attribute field
  36 has a sane value before using it to mask the set of usable attributes.
  37
  38 Reported-by: James Hilliard <james.hilliard1@gmail.com>
  39 Tested-by: James Hilliard <james.hilliard1@gmail.com>
  40 Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
  41 Cc: <stable@vger.kernel.org> # v4.19+
  42 Cc: Borislav Petkov <bp@alien8.de>
  43 Cc: James Morse <james.morse@arm.com>
  44 Cc: Linus Torvalds <torvalds@linux-foundation.org>
  45 Cc: Matt Fleming <matt@codeblueprint.co.uk>
  46 Cc: Peter Jones <pjones@redhat.com>
  47 Cc: Peter Zijlstra <peterz@infradead.org>
  48 Cc: Thomas Gleixner <tglx@linutronix.de>
  49 Cc: linux-efi@vger.kernel.org
  50 Fixes: 38ac0287b7f4 ("fbdev/efifb: Honour UEFI memory map attributes when ...")
  51 Link: http://lkml.kernel.org/r/20190516213159.3530-2-ard.biesheuvel@linaro.org
  52 Signed-off-by: Ingo Molnar <mingo@kernel.org>
  53 Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
  54
  55 ---
  56  drivers/video/fbdev/efifb.c |    8 ++++++--
  57  1 file changed, 6 insertions(+), 2 deletions(-)
  58
  59 --- a/drivers/video/fbdev/efifb.c
  60 +++ b/drivers/video/fbdev/efifb.c
  61 @@ -476,8 +476,12 @@ static int efifb_probe(struct platform_d
  62                  * If the UEFI memory map covers the efifb region, we may only
  63                  * remap it using the attributes the memory map prescribes.
  64                  */
  65 -               mem_flags |= EFI_MEMORY_WT | EFI_MEMORY_WB;
  66 -               mem_flags &= md.attribute;
  67 +               md.attribute &= EFI_MEMORY_UC | EFI_MEMORY_WC |
  68 +                               EFI_MEMORY_WT | EFI_MEMORY_WB;
  69 +               if (md.attribute) {
  70 +                       mem_flags |= EFI_MEMORY_WT | EFI_MEMORY_WB;
  71 +                       mem_flags &= md.attribute;
  72 +               }
  73         }
  74         if (mem_flags & EFI_MEMORY_WC)
  75                 info->screen_base = ioremap_wc(efifb_fix.smem_start,