Skip to contentSkip to conversation

powerpc: Fix kernel-jump address for ppc64 wrapper boot

PATCH In review He Ying 3 participants 6 messages Jun 4, 2021 – Jun 8, 2021 PowerPC

lore
Share & tools
lore

He Ying

author
#0 lore raw

PATCH 1/1 powerpc: Fix kernel-jump address for ppc64 wrapper boot

From "64-bit PowerPC ELF Application Binary Interface Supplement 1.9", we know that the value of a function pointer in a language like C is the address of the function descriptor and the first doubleword of the function descriptor contains the address of the entry point of the function.

So, when we want to jump to an address (e.g. addr) to execute for PPC-elf64abi, we should assign the address of addr *NOT* addr itself to the function pointer or system will jump to the wrong address.

---
 arch/powerpc/boot/main.c | 9 +++++++++
 1 file changed, 9 insertions(+)
arch/powerpc/boot/main.c +9 -0
index cae31a6e8f02..50fd7f11b642 100644
@@ -268,7 +268,16 @@ void start(void)
 	if (console_ops.close)
 		console_ops.close();
 
+#ifdef CONFIG_PPC64_BOOT_WRAPPER
+	/*
+	 * For PPC-elf64abi, the value of a function pointer is the address
+	 * of the function descriptor. And the first doubleword of a function
+	 * descriptor contains the address of the entry point of the function.
+	 */
+	kentry = (kernel_entry_t) &vmlinux.addr;
+#else
 	kentry = (kernel_entry_t) vmlinux.addr;
+#endif
 	if (ft_addr) {
 		if(platform_ops.kentry)
 			platform_ops.kentry(ft_addr, vmlinux.addr);
Signature · 1 line
2.17.1
#1 lore raw

Le 04/06/2021 à 11:22, He Ying a écrit :

Quoted 21 lines

From "64-bit PowerPC ELF Application Binary Interface Supplement 1.9",
we know that the value of a function pointer in a language like C is the address of the function descriptor and the first doubleword of the function descriptor contains the address of the entry point of the function.

So, when we want to jump to an address (e.g. addr) to execute for PPC-elf64abi, we should assign the address of addr *NOT* addr itself to the function pointer or system will jump to the wrong address.

---
  arch/powerpc/boot/main.c | 9 +++++++++
  1 file changed, 9 insertions(+)
arch/powerpc/boot/main.c +0 -0
index cae31a6e8f02..50fd7f11b642 100644

arch/powerpc/boot/main.c void start(void) · line 271

Quoted 4 lines
arch/powerpc/boot/main.c +1 -0
@@ -268,7 +268,16 @@ void start(void)
  	if (console_ops.close)
  		console_ops.close();
 
+#ifdef CONFIG_PPC64_BOOT_WRAPPER

This kind of need doesn't desserve a #ifdef, see https://www.kernel.org/doc/html/latest/process/coding-style.html#conditional-compilation

You can do:

	kentry = (kernel_entry_t)(IS_ENABLED(CONFIG_PPC64_BOOT_WRAPPER) ? &vmlinux.addr : vmlinux.addr);

Or, if you prefer something less compact:

	if (IS_ENABLED(CONFIG_PPC64_BOOT_WRAPPER))
		kentry = (kernel_entry_t) &vmlinux.addr;
	else
		kentry = (kernel_entry_t) vmlinux.addr;
Quoted 13 lines
arch/powerpc/boot/main.c +8 -0
+	/*
+	 * For PPC-elf64abi, the value of a function pointer is the address
+	 * of the function descriptor. And the first doubleword of a function
+	 * descriptor contains the address of the entry point of the function.
+	 */
+	kentry = (kernel_entry_t) &vmlinux.addr;
+#else
  	kentry = (kernel_entry_t) vmlinux.addr;
+#endif
  	if (ft_addr) {
  		if(platform_ops.kentry)
  			platform_ops.kentry(ft_addr, vmlinux.addr);
author
#2 lore raw

Hello,

Quoted 36 lines · Christophe Leroy

Le 04/06/2021 à 11:22, He Ying a écrit :

Quoted 23 lines

 From "64-bit PowerPC ELF Application Binary Interface Supplement 1.9", we know that the value of a function pointer in a language like C is the address of the function descriptor and the first doubleword of the function descriptor contains the address of the entry point of the function.

So, when we want to jump to an address (e.g. addr) to execute for PPC-elf64abi, we should assign the address of addr *NOT* addr itself to the function pointer or system will jump to the wrong address.

---
  arch/powerpc/boot/main.c | 9 +++++++++

  1 file changed, 9 insertions(+)
arch/powerpc/boot/main.c +0 -0
index cae31a6e8f02..50fd7f11b642 100644
@@ -268,7 +268,16 @@ void start(void)
       if (console_ops.close)
           console_ops.close();
Quoted 1 line

  +#ifdef CONFIG_PPC64_BOOT_WRAPPER

This kind of need doesn't desserve a #ifdef, see https://www.kernel.org/doc/html/latest/process/coding-style.html#conditional-compilation

You can do:

    kentry = (kernel_entry_t)(IS_ENABLED(CONFIG_PPC64_BOOT_WRAPPER) ? &vmlinux.addr : vmlinux.addr);

Quoted 7 lines

Or, if you prefer something less compact:

    if (IS_ENABLED(CONFIG_PPC64_BOOT_WRAPPER))
        kentry = (kernel_entry_t) &vmlinux.addr;
    else
        kentry = (kernel_entry_t) vmlinux.addr;

Thanks for reviewing. But from Oliver's reply, this patch should be dropped.

Because all ppc platforms will not build wrapper to ppc64be ELF currently.

And ppc64le uses LE ABI (ABIv2) which doesn't use function descriptors.

So this may not be a problem for now.

Thanks again.

Quoted 18 lines
Quoted 17 lines

+    /*
+     * For PPC-elf64abi, the value of a function pointer is the address +     * of the function descriptor. And the first doubleword of a function
+     * descriptor contains the address of the entry point of the function.
+     */
+    kentry = (kernel_entry_t) &vmlinux.addr;
+#else
      kentry = (kernel_entry_t) vmlinux.addr;
+#endif
      if (ft_addr) {
          if(platform_ops.kentry) platform_ops.kentry(ft_addr, vmlinux.addr);

.

#3 lore raw
In reply to He Ying
Quoted 10 lines · He Ying

From "64-bit PowerPC ELF Application Binary Interface Supplement 1.9", we know that the value of a function pointer in a language like C is the address of the function descriptor and the first doubleword of the function descriptor contains the address of the entry point of the function.

So, when we want to jump to an address (e.g. addr) to execute for PPC-elf64abi, we should assign the address of addr *NOT* addr itself to the function pointer or system will jump to the wrong address.

How have you tested this?

IIRC the 64bit wrapper is only used for ppc64le builds. For that case the current code is work because the LE ABI (ABIv2) doesn't use function descriptors. I think even for a BE kernel we need the current behaviour because the vmlinux's entry point is screwed up (i.e. doesn't point a descriptor) and tools in the wild (probably kexec) expect it to be screwed up.

ABIv2 (LE) reference:
https://openpowerfoundation.org/?resource_lib=64-bit-elf-v2-abi-specification-power-architecture

author
#4 lore raw

Hello,

Quoted 11 lines · Oliver O'Halloran
Quoted 9 lines · He Ying

From "64-bit PowerPC ELF Application Binary Interface Supplement 1.9",
we know that the value of a function pointer in a language like C is the address of the function descriptor and the first doubleword of the function descriptor contains the address of the entry point of the function.

So, when we want to jump to an address (e.g. addr) to execute for PPC-elf64abi, we should assign the address of addr *NOT* addr itself to the function pointer or system will jump to the wrong address.

How have you tested this?

I tested ppc64-elf big-endian. I changed the Kconfig so that ppc64 big-endian

selects PPC64_WRAPPER_BOOT. I used qemu to run the cuImage and found

the problem. It made me confused. By applying this patch, I found it works.

I thought it works for ppc64le too. So I upstream this patch.

Quoted 7 lines

IIRC the 64bit wrapper is only used for ppc64le builds. For that case the current code is work because the LE ABI (ABIv2) doesn't use function descriptors. I think even for a BE kernel we need the current behaviour because the vmlinux's entry point is screwed up (i.e. doesn't point a descriptor) and tools in the wild (probably kexec) expect it to be screwed up.

Yes, you're right. PPC64_WRAPPER_BOOT is only used for ppc64le builds currently.

LE ABI (ABI v2) doesn't use function descriptors. Is that right? I don't test that. If so,

this patch should be dropped. But why does ppc64 have different ABIs? So strange.

If the wrapper is built to ppc64be, my patch is tested right. The entry point in the ELF

header is always right so you can assign the header->e_entry to the function pointer

and then jump to the entry by calling the function. But in the ppc  wrapper, the address

is intialized to 0 or malloced to be an address later. In this situation, I think my patch

should be right for ppc64be.

Quoted 4 lines

ABIv2 (LE) reference:
https://openpowerfoundation.org/?resource_lib=64-bit-elf-v2-abi-specification-power-architecture
.

#5 lore raw
Quoted 33 lines · He Ying

Hello,

Quoted 11 lines · Oliver O'Halloran
Quoted 9 lines · He Ying

From "64-bit PowerPC ELF Application Binary Interface Supplement 1.9",
we know that the value of a function pointer in a language like C is the address of the function descriptor and the first doubleword of the function descriptor contains the address of the entry point of the function.

So, when we want to jump to an address (e.g. addr) to execute for PPC-elf64abi, we should assign the address of addr *NOT* addr itself to the function pointer or system will jump to the wrong address.

How have you tested this?

I tested ppc64-elf big-endian. I changed the Kconfig so that ppc64 big-endian

selects PPC64_WRAPPER_BOOT. I used qemu to run the cuImage and found

the problem. It made me confused. By applying this patch, I found it works.

I thought it works for ppc64le too. So I upstream this patch.

Quoted 7 lines

IIRC the 64bit wrapper is only used for ppc64le builds. For that case the current code is work because the LE ABI (ABIv2) doesn't use function descriptors. I think even for a BE kernel we need the current behaviour because the vmlinux's entry point is screwed up (i.e. doesn't point a descriptor) and tools in the wild (probably kexec) expect it to be screwed up.

Quoted 8 lines

Yes, you're right. PPC64_WRAPPER_BOOT is only used for ppc64le builds currently.

LE ABI (ABI v2) doesn't use function descriptors. Is that right? I don't test that. If so,

this patch should be dropped. But why does ppc64 have different ABIs? So strange.

Yeah, it is strange. When LE support was added the toolchain team took the opportunity to revamp the ABI since BE and LE binaries were never going to be compatible. IIRC there is a slight performance advantage to using v2 since function descriptors added an extra load when performing a non-local function call. I think.

Quoted 6 lines

If the wrapper is built to ppc64be, my patch is tested right. The entry point in the ELF

header is always right so you can assign the header->e_entry to the function pointer

Quoted 7 lines

and then jump to the entry by calling the function. But in the ppc wrapper, the address

is intialized to 0 or malloced to be an address later. In this situation, I think my patch

should be right for ppc64be.

Yeah maybe it's fine. I just have some memories of running into some bizzare edge case at some point. It might have been the entrypoint of the zImage rather than the vmlinux which had (has?) that problem.