Skip to contentSkip to conversation

ARMv8 targets are broken on 2015.04-rc3

U-Boot York Sun 2 participants 6 messages Mar 5, 2015 – Mar 6, 2015 Arm (32-bit)

lore
Share & tools
lore

York Sun

author
#0 lore raw

Guys,

I found these targets are broken on tag 2015.04-rc3

ls2085a_emu_D4
vexpress_aemv8a
vexpress_aemv8a_juno
ls2085a_emu
vexpress_aemv8a_semi
xilinx_zynqmp
ls2085a_simu

git bisect points to this commit a389531439a7d5cea2829054edcf438dc76e79a9. However it is really caused by this one

commit e771a3d538a4fbe235864061ff3c81a8acb11082

Author:     Marc Zyngier <[email protected]>
AuthorDate: Sat Jul 12 14:24:07 2014 +0100
Commit:     Albert ARIBAUD <[email protected]>
CommitDate: Mon Jul 28 17:19:52 2014 +0200

ARM: HYP/non-sec/PSCI: emit DT nodes

Generate the PSCI node in the device tree.

Also add a reserve section for the "secure" code that lives in in normal RAM, so that the kernel knows it'd better not trip on it.

This commit add armv7.h to bootm-fdt.c file.

arch/arm/lib/bootm-fdt.c +1 -0
@@ -17,13 +17,14 @@
 
 #include <common.h>
 #include <fdt_support.h>
+#include <asm/armv7.h>
 

Shouldn't this line be architecture dependent?

York

#1 lore raw
Quoted 42 lines · York Sun

Guys,

I found these targets are broken on tag 2015.04-rc3

ls2085a_emu_D4
vexpress_aemv8a
vexpress_aemv8a_juno
ls2085a_emu
vexpress_aemv8a_semi
xilinx_zynqmp
ls2085a_simu

git bisect points to this commit a389531439a7d5cea2829054edcf438dc76e79a9. However it is really caused by this one

commit e771a3d538a4fbe235864061ff3c81a8acb11082

Author:     Marc Zyngier <[email protected]>
AuthorDate: Sat Jul 12 14:24:07 2014 +0100
Commit:     Albert ARIBAUD <[email protected]>
CommitDate: Mon Jul 28 17:19:52 2014 +0200

ARM: HYP/non-sec/PSCI: emit DT nodes

Generate the PSCI node in the device tree.

Also add a reserve section for the "secure" code that lives in in normal RAM, so that the kernel knows it'd better not trip on it.

This commit add armv7.h to bootm-fdt.c file.

arch/arm/lib/bootm-fdt.c +1 -0
@@ -17,13 +17,14 @@
 
 #include <common.h>
 #include <fdt_support.h>
+#include <asm/armv7.h>
 
Quoted 1 line

Shouldn't this line be architecture dependent?

Looking at the whole commit, how is this breaking things?

Signature · 8 lines
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20150305/00c1effc/attachment.sig>
author
#2 lore raw
Quoted 39 lines · Tom Rini
Quoted 38 lines · York Sun

Guys,

I found these targets are broken on tag 2015.04-rc3

ls2085a_emu_D4
vexpress_aemv8a
vexpress_aemv8a_juno
ls2085a_emu
vexpress_aemv8a_semi
xilinx_zynqmp
ls2085a_simu

git bisect points to this commit a389531439a7d5cea2829054edcf438dc76e79a9. However it is really caused by this one

commit e771a3d538a4fbe235864061ff3c81a8acb11082

Author:     Marc Zyngier <[email protected]>
AuthorDate: Sat Jul 12 14:24:07 2014 +0100
Commit:     Albert ARIBAUD <[email protected]>
CommitDate: Mon Jul 28 17:19:52 2014 +0200

ARM: HYP/non-sec/PSCI: emit DT nodes

Generate the PSCI node in the device tree.

Also add a reserve section for the "secure" code that lives in in normal RAM, so that the kernel knows it'd better not trip on it.

This commit add armv7.h to bootm-fdt.c file.

arch/arm/lib/bootm-fdt.c +0 -0
@@ -17,13 +17,14 @@
 
Quoted 8 lines
Quoted 5 lines
arch/arm/lib/bootm-fdt.c +1 -0
 #include <common.h>
 #include <fdt_support.h>
+#include <asm/armv7.h>
 

Shouldn't this line be architecture dependent?

Looking at the whole commit, how is this breaking things?

Shall this include armv7.h regardless which architecture is compiled? This didn't cause any problem until a recent commit a389531 changes armv7.h.

York

#3 lore raw
Quoted 46 lines · York Sun
Quoted 43 lines · Tom Rini
Quoted 42 lines · York Sun

Guys,

I found these targets are broken on tag 2015.04-rc3

ls2085a_emu_D4
vexpress_aemv8a
vexpress_aemv8a_juno
ls2085a_emu
vexpress_aemv8a_semi
xilinx_zynqmp
ls2085a_simu

git bisect points to this commit a389531439a7d5cea2829054edcf438dc76e79a9. However it is really caused by this one

commit e771a3d538a4fbe235864061ff3c81a8acb11082

Author:     Marc Zyngier <[email protected]>
AuthorDate: Sat Jul 12 14:24:07 2014 +0100
Commit:     Albert ARIBAUD <[email protected]>
CommitDate: Mon Jul 28 17:19:52 2014 +0200

ARM: HYP/non-sec/PSCI: emit DT nodes

Generate the PSCI node in the device tree.

Also add a reserve section for the "secure" code that lives in in normal RAM, so that the kernel knows it'd better not trip on it.

This commit add armv7.h to bootm-fdt.c file.

arch/arm/lib/bootm-fdt.c +1 -0
@@ -17,13 +17,14 @@
 
 #include <common.h>
 #include <fdt_support.h>
+#include <asm/armv7.h>
 
Quoted 6 lines
Quoted 3 lines
Quoted 1 line

Shouldn't this line be architecture dependent?

Looking at the whole commit, how is this breaking things?

Shall this include armv7.h regardless which architecture is compiled? This didn't cause any problem until a recent commit a389531 changes armv7.h.

Yes, the include looks safe. This is one of those cryptic bisects, can you poke it a bit more and see what's causing the failure inside of the file?

Signature · 8 lines
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20150305/8ebb82c9/attachment.sig>
author
#4 lore raw
Quoted 51 lines · Tom Rini
Quoted 50 lines · York Sun
Quoted 46 lines · Tom Rini
Quoted 42 lines · York Sun

Guys,

I found these targets are broken on tag 2015.04-rc3

ls2085a_emu_D4
vexpress_aemv8a
vexpress_aemv8a_juno
ls2085a_emu
vexpress_aemv8a_semi
xilinx_zynqmp
ls2085a_simu

git bisect points to this commit a389531439a7d5cea2829054edcf438dc76e79a9. However it is really caused by this one

commit e771a3d538a4fbe235864061ff3c81a8acb11082

Author:     Marc Zyngier <[email protected]>
AuthorDate: Sat Jul 12 14:24:07 2014 +0100
Commit:     Albert ARIBAUD <[email protected]>
CommitDate: Mon Jul 28 17:19:52 2014 +0200

ARM: HYP/non-sec/PSCI: emit DT nodes

Generate the PSCI node in the device tree.

Also add a reserve section for the "secure" code that lives in in normal RAM, so that the kernel knows it'd better not trip on it.

This commit add armv7.h to bootm-fdt.c file.

arch/arm/lib/bootm-fdt.c +1 -0
@@ -17,13 +17,14 @@
 
 #include <common.h>
 #include <fdt_support.h>
+#include <asm/armv7.h>
 
Quoted 1 line

Shouldn't this line be architecture dependent?

Looking at the whole commit, how is this breaking things?

Quoted 7 lines
Quoted 2 lines

Shall this include armv7.h regardless which architecture is compiled? This didn't cause any problem until a recent commit a389531 changes armv7.h.

Yes, the include looks safe. This is one of those cryptic bisects, can you poke it a bit more and see what's causing the failure inside of the file?

You mean it is safe to include armv7.h even if I am building armv8 targets? When compiling armv8 targets, I have these warnings

/home/u-boot-upstream-85xx/arch/arm/include/asm/armv7.h: In function ?v7_enable_smp?:
/home/u-boot-upstream-85xx/arch/arm/include/asm/io.h:72:28: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]

 #define __arch_getl(a)   (*(volatile unsigned int *)(a))
                            ^
/home/u-boot-upstream-85xx/arch/arm/include/asm/io.h:151:31: note: in expansion
of macro ?__arch_getl?
 #define readl(c) ({ u32 __v = __arch_getl(c); __iormb(); __v; })
                               ^
/home/u-boot-upstream-85xx/arch/arm/include/asm/armv7.h:105:9: note: in
expansion of macro ?readl?
  temp = readl(address);
         ^

The readl was added into armv7.h by commit a389531.

York

#5 lore raw
Quoted 79 lines · York Sun
Quoted 51 lines · Tom Rini
Quoted 50 lines · York Sun
Quoted 46 lines · Tom Rini
Quoted 42 lines · York Sun

Guys,

I found these targets are broken on tag 2015.04-rc3

ls2085a_emu_D4
vexpress_aemv8a
vexpress_aemv8a_juno
ls2085a_emu
vexpress_aemv8a_semi
xilinx_zynqmp
ls2085a_simu

git bisect points to this commit a389531439a7d5cea2829054edcf438dc76e79a9. However it is really caused by this one

commit e771a3d538a4fbe235864061ff3c81a8acb11082

Author:     Marc Zyngier <[email protected]>
AuthorDate: Sat Jul 12 14:24:07 2014 +0100
Commit:     Albert ARIBAUD <[email protected]>
CommitDate: Mon Jul 28 17:19:52 2014 +0200

ARM: HYP/non-sec/PSCI: emit DT nodes

Generate the PSCI node in the device tree.

Also add a reserve section for the "secure" code that lives in in normal RAM, so that the kernel knows it'd better not trip on it.

This commit add armv7.h to bootm-fdt.c file.

arch/arm/lib/bootm-fdt.c +1 -0
@@ -17,13 +17,14 @@
 
 #include <common.h>
 #include <fdt_support.h>
+#include <asm/armv7.h>
 
Quoted 1 line

Shouldn't this line be architecture dependent?

Looking at the whole commit, how is this breaking things?

Quoted 6 lines
Quoted 2 lines

Shall this include armv7.h regardless which architecture is compiled? This didn't cause any problem until a recent commit a389531 changes armv7.h.

Yes, the include looks safe. This is one of those cryptic bisects, can you poke it a bit more and see what's causing the failure inside of the file?

You mean it is safe to include armv7.h even if I am building armv8 targets? When compiling armv8 targets, I have these warnings

/home/u-boot-upstream-85xx/arch/arm/include/asm/armv7.h: In function ?v7_enable_smp?:
/home/u-boot-upstream-85xx/arch/arm/include/asm/io.h:72:28: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]

 #define __arch_getl(a)   (*(volatile unsigned int *)(a))
                            ^
/home/u-boot-upstream-85xx/arch/arm/include/asm/io.h:151:31: note: in expansion
of macro ?__arch_getl?
 #define readl(c) ({ u32 __v = __arch_getl(c); __iormb(); __v; })
                               ^
/home/u-boot-upstream-85xx/arch/arm/include/asm/armv7.h:105:9: note: in
expansion of macro ?readl?
  temp = readl(address);
         ^
Quoted 1 line

The readl was added into armv7.h by commit a389531.

Ah well now we're getting somewhere :) Let me try and whip up a patch..

Signature · 8 lines
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20150305/a88caf5d/attachment.sig>