mbox series

[for-4.1?,0/2] arm: further improve initrd positioning

Message ID 20190722151804.25467-1-peter.maydell@linaro.org
Headers show
Series arm: further improve initrd positioning | expand

Message

Peter Maydell July 22, 2019, 3:18 p.m. UTC
In commit e6b2b20d9735d4ef we made the boot loader code try to avoid
putting the initrd on top of the kernel.  However the expression used
to calculate the start of the initrd:

    info->initrd_start = info->loader_start +
        MAX(MIN(info->ram_size / 2, 128 * 1024 * 1024), kernel_size);

incorrectly uses 'kernel_size' as the offset within RAM of the
highest address to avoid.  This is incorrect because the kernel
doesn't start at address 0, but slightly higher than that.  This
means that we can still incorrectly end up overlaying the initrd on
the kernel in some cases, for example:

* The kernel's image_size is 0x0a7a8000
* The kernel was loaded at   0x40080000
* The end of the kernel is   0x4A828000
* The DTB was loaded at      0x4a800000

To get this right we need to track the actual highest address used
by the kernel and use that rather than kernel_size. We already
trace the low_addr and high_addr for ELF images; set them
also for the various other image types we support, and then use
high_addr as the lowest allowed address for the initrd.

Patch 1 just does a preliminary variable rename; patch 2 is the meat.

Only very lightly tested...

Marked as 'maybe for 4.1' because it is a bug fix and to code which
is new in 4.1. OTOH cases that fail now would have failed with 4.0
so it is not a regression strictly speaking. And we're getting
steadily closer to release and I haven't very heavily tested this
patch. I incline towards including it, overall.

thanks
-- PMM

Peter Maydell (2):
  hw/arm/boot: Rename elf_{low,high}_addr to image_{low,high}_addr
  hw/arm/boot: Further improve initrd positioning code

 hw/arm/boot.c | 37 +++++++++++++++++++++++++++----------
 1 file changed, 27 insertions(+), 10 deletions(-)

-- 
2.20.1

Comments

Mark Rutland July 22, 2019, 4:52 p.m. UTC | #1
Hi Peter,

On Mon, Jul 22, 2019 at 04:18:02PM +0100, Peter Maydell wrote:
> In commit e6b2b20d9735d4ef we made the boot loader code try to avoid

> putting the initrd on top of the kernel.  However the expression used

> to calculate the start of the initrd:

> 

>     info->initrd_start = info->loader_start +

>         MAX(MIN(info->ram_size / 2, 128 * 1024 * 1024), kernel_size);

> 

> incorrectly uses 'kernel_size' as the offset within RAM of the

> highest address to avoid.  This is incorrect because the kernel

> doesn't start at address 0, but slightly higher than that.  This

> means that we can still incorrectly end up overlaying the initrd on

> the kernel in some cases, for example:

> 

> * The kernel's image_size is 0x0a7a8000

> * The kernel was loaded at   0x40080000

> * The end of the kernel is   0x4A828000

> * The DTB was loaded at      0x4a800000

> 

> To get this right we need to track the actual highest address used

> by the kernel and use that rather than kernel_size. We already

> trace the low_addr and high_addr for ELF images; set them

> also for the various other image types we support, and then use

> high_addr as the lowest allowed address for the initrd.

> 

> Patch 1 just does a preliminary variable rename; patch 2 is the meat.

> 

> Only very lightly tested...


FWIW, I've given this a spin, and it manages to boot my growing
collection of problematic images, along with all the previously-working
images I had lying around.

For both patches, feel free to take that as:

Tested-by: Mark Rutland <mark.rutland@arm.com>


> Marked as 'maybe for 4.1' because it is a bug fix and to code which

> is new in 4.1. OTOH cases that fail now would have failed with 4.0

> so it is not a regression strictly speaking. And we're getting

> steadily closer to release and I haven't very heavily tested this

> patch. I incline towards including it, overall.


Having these in 4.1 would be nice. AIUI the logic is correct, and this
seems low risk to me, but it is your call.

If you decide to hold off for a release, I can apply these patches
locally without too much pain.

Thanks for dealing with this, anyhow!

Mark.

> 

> thanks

> -- PMM

> 

> Peter Maydell (2):

>   hw/arm/boot: Rename elf_{low,high}_addr to image_{low,high}_addr

>   hw/arm/boot: Further improve initrd positioning code

> 

>  hw/arm/boot.c | 37 +++++++++++++++++++++++++++----------

>  1 file changed, 27 insertions(+), 10 deletions(-)

> 

> -- 

> 2.20.1

>