[2/2] ARM: qemu-arm: define fdt_addr_r

Message ID 20181012050757.6925-2-takahiro.akashi@linaro.org
State New
Headers show
Series
  • [1/2] efi_loader: rework fdt handling in distro boot script
Related show

Commit Message

AKASHI Takahiro Oct. 12, 2018, 5:07 a.m.
This variable, fdt_addr_t, is missing in the current qemu-arm.h while it
seems to be mandatory, at least, to run distro_bootcmd as expected.
So just add its definition. A size of 1MB would be enough.

Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
---
 include/configs/qemu-arm.h | 1 +
 1 file changed, 1 insertion(+)

Comments

Tuomas Tynkkynen Oct. 15, 2018, 1:01 a.m. | #1
Hi Takahiro,

On Fri, 12 Oct 2018 14:07:57 +0900
AKASHI Takahiro <takahiro.akashi@linaro.org> wrote:

> This variable, fdt_addr_t, is missing in the current qemu-arm.h while
> it seems to be mandatory, at least, to run distro_bootcmd as expected.
> So just add its definition. A size of 1MB would be enough.
> 

In what way is this required for distro_bootcmd to work? At least in the
past I've tested qemu_arm64_defconfig and EFI boot with the Fedora
netinst image and it has worked fine in stock U-Boot.

Note that these '-machine virt' based targets are slightly different
from real hardware in the sense that instead of loading a .dtb file
provided by the OS, the device tree is provided by QEMU. In the hunk
below you can see "fdt_addr=0x40000000\0" providing the address of
the QEMU-provided device tree which the distro scripts should be
using.

I guess in principle having ${fdt_addr_r} set as well shouldn't hurt and
might be used for testing/unusual purposes. Glancing at cmd/pxe.c,
there is a problem though, in that if ${fdt_addr_r} were defined, a PXE
file using the FDTDIR directive would attempt loading a dtb file named
"<NULL>-qemu-arm.dtb" instead of falling back to using ${fdt_addr}.
That bug would need to be fixed first before applying this patch.

> Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
> ---
>  include/configs/qemu-arm.h | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/include/configs/qemu-arm.h b/include/configs/qemu-arm.h
> index 91fb8d47edf8..0e66f946dde5 100644
> --- a/include/configs/qemu-arm.h
> +++ b/include/configs/qemu-arm.h
> @@ -55,6 +55,7 @@
>  	"fdt_high=0xffffffff\0" \
>  	"initrd_high=0xffffffff\0" \
>  	"fdt_addr=0x40000000\0" \
> +	"fdt_addr_r=0x40100000\0" \
>  	"scriptaddr=0x40200000\0" \
>  	"pxefile_addr_r=0x40300000\0" \
>  	"kernel_addr_r=0x40400000\0" \
AKASHI Takahiro Oct. 15, 2018, 5:14 a.m. | #2
On Mon, Oct 15, 2018 at 04:01:06AM +0300, Tuomas Tynkkynen wrote:
> Hi Takahiro,
> 
> On Fri, 12 Oct 2018 14:07:57 +0900
> AKASHI Takahiro <takahiro.akashi@linaro.org> wrote:
> 
> > This variable, fdt_addr_t, is missing in the current qemu-arm.h while
> > it seems to be mandatory, at least, to run distro_bootcmd as expected.
> > So just add its definition. A size of 1MB would be enough.
> > 
> 
> In what way is this required for distro_bootcmd to work? At least in the
> past I've tested qemu_arm64_defconfig and EFI boot with the Fedora
> netinst image and it has worked fine in stock U-Boot.
> 
> Note that these '-machine virt' based targets are slightly different
> from real hardware in the sense that instead of loading a .dtb file
> provided by the OS, the device tree is provided by QEMU. In the hunk
> below you can see "fdt_addr=0x40000000\0" providing the address of
> the QEMU-provided device tree which the distro scripts should be
> using.

Yep, I know that.
I was not clear; what distro bootcmd, or more specifically scan_dev_for_efi,
tries to do regarding fdt handling is that, if "fdtfile" variable is defined,
it supersedes qemu's own dtb (that is what "fdt_addr" points to), but
this trick doesn't work expectedly if "fdt_addr_r" variable is not defined.

> I guess in principle having ${fdt_addr_r} set as well shouldn't hurt and
> might be used for testing/unusual purposes.

I don't know whether the case is "unsual" or not.
But doc/README.distro cleary says that fdt_addr_r is *mandatory*.
If u-boot works without it, it's a bug, otherwise we must correct
the doc (or scan_dev_for_efi in the first place?).

Thanks,
-Takahiro Akashi

> Glancing at cmd/pxe.c,
> there is a problem though, in that if ${fdt_addr_r} were defined, a PXE
> file using the FDTDIR directive would attempt loading a dtb file named
> "<NULL>-qemu-arm.dtb" instead of falling back to using ${fdt_addr}.
> That bug would need to be fixed first before applying this patch.
>
> > Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
> > ---
> >  include/configs/qemu-arm.h | 1 +
> >  1 file changed, 1 insertion(+)
> > 
> > diff --git a/include/configs/qemu-arm.h b/include/configs/qemu-arm.h
> > index 91fb8d47edf8..0e66f946dde5 100644
> > --- a/include/configs/qemu-arm.h
> > +++ b/include/configs/qemu-arm.h
> > @@ -55,6 +55,7 @@
> >  	"fdt_high=0xffffffff\0" \
> >  	"initrd_high=0xffffffff\0" \
> >  	"fdt_addr=0x40000000\0" \
> > +	"fdt_addr_r=0x40100000\0" \
> >  	"scriptaddr=0x40200000\0" \
> >  	"pxefile_addr_r=0x40300000\0" \
> >  	"kernel_addr_r=0x40400000\0" \
>
Alexander Graf Oct. 16, 2018, 1:04 p.m. | #3
On 15.10.18 07:14, AKASHI Takahiro wrote:
> On Mon, Oct 15, 2018 at 04:01:06AM +0300, Tuomas Tynkkynen wrote:
>> Hi Takahiro,
>>
>> On Fri, 12 Oct 2018 14:07:57 +0900
>> AKASHI Takahiro <takahiro.akashi@linaro.org> wrote:
>>
>>> This variable, fdt_addr_t, is missing in the current qemu-arm.h while
>>> it seems to be mandatory, at least, to run distro_bootcmd as expected.
>>> So just add its definition. A size of 1MB would be enough.
>>>
>>
>> In what way is this required for distro_bootcmd to work? At least in the
>> past I've tested qemu_arm64_defconfig and EFI boot with the Fedora
>> netinst image and it has worked fine in stock U-Boot.
>>
>> Note that these '-machine virt' based targets are slightly different
>> from real hardware in the sense that instead of loading a .dtb file
>> provided by the OS, the device tree is provided by QEMU. In the hunk
>> below you can see "fdt_addr=0x40000000\0" providing the address of
>> the QEMU-provided device tree which the distro scripts should be
>> using.
> 
> Yep, I know that.
> I was not clear; what distro bootcmd, or more specifically scan_dev_for_efi,
> tries to do regarding fdt handling is that, if "fdtfile" variable is defined,
> it supersedes qemu's own dtb (that is what "fdt_addr" points to), but
> this trick doesn't work expectedly if "fdt_addr_r" variable is not defined.
> 
>> I guess in principle having ${fdt_addr_r} set as well shouldn't hurt and
>> might be used for testing/unusual purposes.
> 
> I don't know whether the case is "unsual" or not.
> But doc/README.distro cleary says that fdt_addr_r is *mandatory*.
> If u-boot works without it, it's a bug, otherwise we must correct
> the doc (or scan_dev_for_efi in the first place?).

I agree. Boards that support distro boot *must* provide fdt_addr_r.
Otherwise distro scripts may fail unexpectedly because they assume the
variable is present.

> 
> Thanks,
> -Takahiro Akashi
> 
>> Glancing at cmd/pxe.c,
>> there is a problem though, in that if ${fdt_addr_r} were defined, a PXE
>> file using the FDTDIR directive would attempt loading a dtb file named
>> "<NULL>-qemu-arm.dtb" instead of falling back to using ${fdt_addr}.
>> That bug would need to be fixed first before applying this patch.

Well, and that load will fail and everyone's happy, no? IMHO we should
fall back to $fdtcontroladdr always, but the pxe boot path is not
something I would endorse onto anyone (what you want as PXE really is
called 'dhcp' in the efi_loader distro boot world)


Alex
Tuomas Tynkkynen Oct. 17, 2018, 10:25 p.m. | #4
Hi Alexander,

On Tue, 16 Oct 2018 15:04:26 +0200
Alexander Graf <agraf@suse.de> wrote:

...
> >   
> >> Glancing at cmd/pxe.c,
> >> there is a problem though, in that if ${fdt_addr_r} were defined,
> >> a PXE file using the FDTDIR directive would attempt loading a dtb
> >> file named "<NULL>-qemu-arm.dtb" instead of falling back to using
> >> ${fdt_addr}. That bug would need to be fixed first before applying
> >> this patch.  
> 
> Well, and that load will fail and everyone's happy, no? 

No, because currently if DTB loading from the filesystem/tftp is
attempted and it fails, it aborts the boot. It doesn't matter if it's
via a 'FDT' or 'FDTDIR' directive. In the case of typical hardware
that's probably the desired behaviour.

I guess the qemu_arm + FDTDIR case can be fixed by not attempting
a .dtb load if the default fdtfile is not known due to $soc or $board
being unset. At least I doubt that some other board could be relying
on a filename containing literally "<NULL>" :)

> IMHO we should
> fall back to $fdtcontroladdr always

FWIW, to me the idea of passing $fdtcontroladdr to the OS has always
filled me with dread. On top of the usual backwards- and
forwards-compatibility problems that happen when mixing and matching
kernels and DTBs from different releases, you now have to deal with
issues like U-Boot specific .dts that are majorly diverged from Linux
ones, or where the .dts is otherwise from Linux but the U-Boot specific
additions break it for Linux, or where the .dts used is wrong for the
specific hardware revision but close enough for U-Boot's purposes,
and so on...

- Tuomas
Alexander Graf Oct. 18, 2018, 7:25 a.m. | #5
On 18.10.18 00:25, Tuomas Tynkkynen wrote:
> Hi Alexander,
> 
> On Tue, 16 Oct 2018 15:04:26 +0200
> Alexander Graf <agraf@suse.de> wrote:
> 
> ...
>>>   
>>>> Glancing at cmd/pxe.c,
>>>> there is a problem though, in that if ${fdt_addr_r} were defined,
>>>> a PXE file using the FDTDIR directive would attempt loading a dtb
>>>> file named "<NULL>-qemu-arm.dtb" instead of falling back to using
>>>> ${fdt_addr}. That bug would need to be fixed first before applying
>>>> this patch.  
>>
>> Well, and that load will fail and everyone's happy, no? 
> 
> No, because currently if DTB loading from the filesystem/tftp is
> attempted and it fails, it aborts the boot. It doesn't matter if it's
> via a 'FDT' or 'FDTDIR' directive. In the case of typical hardware
> that's probably the desired behaviour.
> 
> I guess the qemu_arm + FDTDIR case can be fixed by not attempting
> a .dtb load if the default fdtfile is not known due to $soc or $board
> being unset. At least I doubt that some other board could be relying
> on a filename containing literally "<NULL>" :)

Well - IIRC $soc and $board should also always be defined ;). So yet
another thing to fix in the QEMU port.

> 
>> IMHO we should
>> fall back to $fdtcontroladdr always
> 
> FWIW, to me the idea of passing $fdtcontroladdr to the OS has always
> filled me with dread. On top of the usual backwards- and
> forwards-compatibility problems that happen when mixing and matching
> kernels and DTBs from different releases, you now have to deal with

That's something that we seriously as a community need to get sorted
out. We're pushing hard for it in the EBBR context. The fact that people
are afraid of putting *hardware desciption* into their firmware is just
mind boggling.

> issues like U-Boot specific .dts that are majorly diverged from Linux
> ones, or where the .dts is otherwise from Linux but the U-Boot specific

These case should really be the minority. And if you see those, please
fix them.

> additions break it for Linux, or where the .dts used is wrong for the

I've never seen a case where a U-Boot addition broke the DT for Linux.

> specific hardware revision but close enough for U-Boot's purposes,
> and so on...

Again, something that just needs fixing. Device trees belong to the
entity that knows hardware, not to the OS. Otherwise you lose the
abstraction layer that DT gives you and you lose the ability to run
"generic" kernels. And of course you break the ecosystem, because now
good luck running BSD, your own little toy OS, etc ;)


Alex
AKASHI Takahiro Oct. 19, 2018, 6:33 a.m. | #6
On Thu, Oct 18, 2018 at 09:25:04AM +0200, Alexander Graf wrote:
> 
> 
> On 18.10.18 00:25, Tuomas Tynkkynen wrote:
> > Hi Alexander,
> > 
> > On Tue, 16 Oct 2018 15:04:26 +0200
> > Alexander Graf <agraf@suse.de> wrote:
> > 
> > ...
> >>>   
> >>>> Glancing at cmd/pxe.c,
> >>>> there is a problem though, in that if ${fdt_addr_r} were defined,
> >>>> a PXE file using the FDTDIR directive would attempt loading a dtb
> >>>> file named "<NULL>-qemu-arm.dtb" instead of falling back to using
> >>>> ${fdt_addr}. That bug would need to be fixed first before applying
> >>>> this patch.  
> >>
> >> Well, and that load will fail and everyone's happy, no? 
> > 
> > No, because currently if DTB loading from the filesystem/tftp is
> > attempted and it fails, it aborts the boot. It doesn't matter if it's
> > via a 'FDT' or 'FDTDIR' directive. In the case of typical hardware
> > that's probably the desired behaviour.
> > 
> > I guess the qemu_arm + FDTDIR case can be fixed by not attempting
> > a .dtb load if the default fdtfile is not known due to $soc or $board
> > being unset. At least I doubt that some other board could be relying
> > on a filename containing literally "<NULL>" :)
> 
> Well - IIRC $soc and $board should also always be defined ;). So yet
> another thing to fix in the QEMU port.

See my patch below. Are you happy with it?
(qemu-qemu-arm.dtb doesn't make sense to me, though :)

> > 
> >> IMHO we should
> >> fall back to $fdtcontroladdr always
> > 
> > FWIW, to me the idea of passing $fdtcontroladdr to the OS has always
> > filled me with dread. On top of the usual backwards- and
> > forwards-compatibility problems that happen when mixing and matching
> > kernels and DTBs from different releases, you now have to deal with
> 
> That's something that we seriously as a community need to get sorted
> out. We're pushing hard for it in the EBBR context. The fact that people
> are afraid of putting *hardware desciption* into their firmware is just
> mind boggling.

I don't have a strong opinion, but it seems to me that 'fall-back' approach
is quite normal. If it doesn't work on a specific board, 'fdt_addr'
should be defined.

Thanks,
-Takahiro Akashi


> > issues like U-Boot specific .dts that are majorly diverged from Linux
> > ones, or where the .dts is otherwise from Linux but the U-Boot specific
> 
> These case should really be the minority. And if you see those, please
> fix them.
> 
> > additions break it for Linux, or where the .dts used is wrong for the
> 
> I've never seen a case where a U-Boot addition broke the DT for Linux.
> 
> > specific hardware revision but close enough for U-Boot's purposes,
> > and so on...
> 
> Again, something that just needs fixing. Device trees belong to the
> entity that knows hardware, not to the OS. Otherwise you lose the
> abstraction layer that DT gives you and you lose the ability to run
> "generic" kernels. And of course you break the ecosystem, because now
> good luck running BSD, your own little toy OS, etc ;)
> 
> 
> Alex

===8<===
From 47b26a86359a3b612e890f2ca2d5d20092f9f4e1 Mon Sep 17 00:00:00 2001
From: AKASHI Takahiro <takahiro.akashi@linaro.org>
Date: Fri, 19 Oct 2018 15:22:02 +0900
Subject: [PATCH] arm: qemu: rework Kconfig

Define SYS_SOC and move SYS_* to a canonical place (under board).

Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
---
 arch/arm/mach-qemu/Kconfig       | 18 ++++++++++--------
 board/emulation/Kconfig          |  2 ++
 board/emulation/qemu-arm/Kconfig |  9 +++++++++
 3 files changed, 21 insertions(+), 8 deletions(-)
 create mode 100644 board/emulation/qemu-arm/Kconfig

diff --git a/arch/arm/mach-qemu/Kconfig b/arch/arm/mach-qemu/Kconfig
index a2e4b98b8887..d75a95183a75 100644
--- a/arch/arm/mach-qemu/Kconfig
+++ b/arch/arm/mach-qemu/Kconfig
@@ -3,22 +3,24 @@ if ARCH_QEMU
 config SYS_VENDOR
 	default "emulation"
 
-config SYS_BOARD
-	default "qemu-arm"
+config SYS_SOC
+	default "qemu"
 
-config SYS_CONFIG_NAME
-	default "qemu-arm"
-
-endif
+choice
+	prompt "QEMU cpu type"
 
 config TARGET_QEMU_ARM_32BIT
-	bool "Support qemu_arm"
+	bool "Arm"
 	depends on ARCH_QEMU
 	select ARCH_SUPPORT_PSCI
 	select CPU_V7A
 	select SYS_ARCH_TIMER
 
 config TARGET_QEMU_ARM_64BIT
-	bool "Support qemu_arm64"
+	bool "AArch64"
 	depends on ARCH_QEMU
 	select ARM64
+
+endchoice
+
+endif
diff --git a/board/emulation/Kconfig b/board/emulation/Kconfig
index f821458fa6ac..597e926cdb11 100644
--- a/board/emulation/Kconfig
+++ b/board/emulation/Kconfig
@@ -27,4 +27,6 @@ endchoice
 
 source "board/emulation/qemu-x86/Kconfig"
 
+source "board/emulation/qemu-arm/Kconfig"
+
 endif
diff --git a/board/emulation/qemu-arm/Kconfig b/board/emulation/qemu-arm/Kconfig
new file mode 100644
index 000000000000..db8b2a4dfae2
--- /dev/null
+++ b/board/emulation/qemu-arm/Kconfig
@@ -0,0 +1,9 @@
+if TARGET_QEMU_ARM_32BIT || TARGET_QEMU_ARM_64BIT
+
+config SYS_BOARD
+	default "qemu-arm"
+
+config SYS_CONFIG_NAME
+	default "qemu-arm"
+
+endif

Patch

diff --git a/include/configs/qemu-arm.h b/include/configs/qemu-arm.h
index 91fb8d47edf8..0e66f946dde5 100644
--- a/include/configs/qemu-arm.h
+++ b/include/configs/qemu-arm.h
@@ -55,6 +55,7 @@ 
 	"fdt_high=0xffffffff\0" \
 	"initrd_high=0xffffffff\0" \
 	"fdt_addr=0x40000000\0" \
+	"fdt_addr_r=0x40100000\0" \
 	"scriptaddr=0x40200000\0" \
 	"pxefile_addr_r=0x40300000\0" \
 	"kernel_addr_r=0x40400000\0" \