Re: [PATCH] x86/mm, efi: Check for valid image type

From: Matt Fleming
Date: Tue Jul 28 2015 - 16:52:07 EST


(Pulling in Josh)

On Wed, 22 Jul, at 05:32:44PM, Sebastian Andrzej Siewior wrote:
> I usually see
> |Ignoring BGRT: failed to allocate memory for image (wanted 264301314 bytes)
> |Ignoring BGRT: failed to allocate memory for image (wanted 3925872891 bytes)
>
> sometimes I get
>
> |------------[ cut here ]------------
> |WARNING: CPU: 0 PID: 0 at mm/early_ioremap.c:136 __early_ioremap.constprop.0+0x113/0x1d3()
> â
> | [<ffffffff81b3de8c>] __early_ioremap.constprop.0+0x113/0x1d3
> | [<ffffffff81b3e106>] early_ioremap+0x13/0x15
> | [<ffffffff81b2c4a9>] efi_bgrt_init+0x1e2/0x27d
> â
>
> now and then. The data behind that pointer changes on each boot because
> nobody preserves the content across kexec.
>
> Signed-off-by: Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx>
> ---
>
> I don't know much about the requirement of having the .bmp in memory all the
> time. Would it be a bad thing to compress the bmp and uncompress on cat from
> userland? In my case the bmp has 272 KiB and LZO gets it down to 12KiB,
> XZ 7.4KiB.

The usual use for BGRT is to display it during kernel boot, so
interacting with userland doesn't help you there.

> arch/x86/platform/efi/efi-bgrt.c | 4 ++++
> 1 file changed, 4 insertions(+)
>
> diff --git a/arch/x86/platform/efi/efi-bgrt.c b/arch/x86/platform/efi/efi-bgrt.c
> index d7f997f7c26d..59710f0875bb 100644
> --- a/arch/x86/platform/efi/efi-bgrt.c
> +++ b/arch/x86/platform/efi/efi-bgrt.c
> @@ -79,6 +79,10 @@ void __init efi_bgrt_init(void)
> memcpy_fromio(&bmp_header, image, sizeof(bmp_header));
> if (ioremapped)
> early_iounmap(image, sizeof(bmp_header));
> + if (bmp_header.id != 0x4d42) {
> + pr_err("BGRT: Not a valid BMP file.\n");
> + return;
> + }
> bgrt_image_size = bmp_header.size;
>
> bgrt_image = kmalloc(bgrt_image_size, GFP_KERNEL | __GFP_NOWARN);

I'm confused, is the BMP image valid on your machine or not? You add a
validity check but talk about compressing it above.

--
Matt Fleming, Intel Open Source Technology Center
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/