[PATCH v1 1/3] Recieve boot_param from RDI Instead of RSI

[PATCH v1 1/3] Recieve boot_param from RDI Instead of RSI

[Khalid Ali]

From: Khalid Ali <khaliidcaliy@gmail.com>

Adjust the kernel entry point to recieve arguments from RDI instead of
RSI.

Signed-off-by: Khalid Ali <khaliidcaliy@gmail.com>
---
 arch/x86/kernel/head_64.S | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/head_64.S b/arch/x86/kernel/head_64.S
index dfb5390e5c9a..d24fea15b6a6 100644
--- a/arch/x86/kernel/head_64.S
+++ b/arch/x86/kernel/head_64.S
@@ -43,7 +43,7 @@ SYM_CODE_START_NOALIGN(startup_64)
 	 * for us.  These identity mapped page tables map all of the
 	 * kernel pages and possibly all of memory.
 	 *
-	 * %RSI holds the physical address of the boot_params structure
+	 * %RDI holds the physical address of the boot_params structure
 	 * provided by the bootloader. Preserve it in %R15 so C function calls
 	 * will not clobber it.
 	 *
@@ -56,7 +56,7 @@ SYM_CODE_START_NOALIGN(startup_64)
 	 * compiled to run at we first fixup the physical addresses in our page
 	 * tables and then reload them.
 	 */
-	mov	%rsi, %r15
+	mov	%rdi, %r15
 
 	/* Set up the stack for verify_cpu() */
 	leaq	__top_init_kernel_stack(%rip), %rsp
-- 
2.49.0

Re: [PATCH v1 1/3] Recieve boot_param from RDI Instead of RSI

[H. Peter Anvin]

On June 23, 2025 1:15:23 AM PDT, Khalid Ali <khaliidcaliy@gmail.com> wrote:
>From: Khalid Ali <khaliidcaliy@gmail.com>
>
>Adjust the kernel entry point to recieve arguments from RDI instead of
>RSI.
>
>Signed-off-by: Khalid Ali <khaliidcaliy@gmail.com>
>---
> arch/x86/kernel/head_64.S | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
>diff --git a/arch/x86/kernel/head_64.S b/arch/x86/kernel/head_64.S
>index dfb5390e5c9a..d24fea15b6a6 100644
>--- a/arch/x86/kernel/head_64.S
>+++ b/arch/x86/kernel/head_64.S
>@@ -43,7 +43,7 @@ SYM_CODE_START_NOALIGN(startup_64)
> 	 * for us.  These identity mapped page tables map all of the
> 	 * kernel pages and possibly all of memory.
> 	 *
>-	 * %RSI holds the physical address of the boot_params structure
>+	 * %RDI holds the physical address of the boot_params structure
> 	 * provided by the bootloader. Preserve it in %R15 so C function calls
> 	 * will not clobber it.
> 	 *
>@@ -56,7 +56,7 @@ SYM_CODE_START_NOALIGN(startup_64)
> 	 * compiled to run at we first fixup the physical addresses in our page
> 	 * tables and then reload them.
> 	 */
>-	mov	%rsi, %r15
>+	mov	%rdi, %r15
> 
> 	/* Set up the stack for verify_cpu() */
> 	leaq	__top_init_kernel_stack(%rip), %rsp

This is an ABI.

Nacked-by: H. Peter Anvin <hpa@zytor.com>

Re:[PATCH v1 1/3] Recieve boot_param from RDI Instead of RSI

[Khalid Ali]

> On June 23, 2025 1:15:23 AM PDT, Khalid Ali <khaliidcaliy@gmail.com> wrote:
> >From: Khalid Ali <khaliidcaliy@gmail.com>
> >
> >Adjust the kernel entry point to recieve arguments from RDI instead of
> >RSI.
> >
> >Signed-off-by: Khalid Ali <khaliidcaliy@gmail.com>
> >---
> > arch/x86/kernel/head_64.S | 4 ++--
> > 1 file changed, 2 insertions(+), 2 deletions(-)
> >
> >diff --git a/arch/x86/kernel/head_64.S b/arch/x86/kernel/head_64.S
> >index dfb5390e5c9a..d24fea15b6a6 100644
> >--- a/arch/x86/kernel/head_64.S
> >+++ b/arch/x86/kernel/head_64.S
> >@@ -43,7 +43,7 @@ SYM_CODE_START_NOALIGN(startup_64)
> >        * for us.  These identity mapped page tables map all of the
> >        * kernel pages and possibly all of memory.
> >        *
> >-       * %RSI holds the physical address of the boot_params structure
> >+       * %RDI holds the physical address of the boot_params structure
> >        * provided by the bootloader. Preserve it in %R15 so C function calls
> >        * will not clobber it.
> >        *
> >@@ -56,7 +56,7 @@ SYM_CODE_START_NOALIGN(startup_64)
> >        * compiled to run at we first fixup the physical addresses in our page
> >        * tables and then reload them.
> >        */
> >-      mov     %rsi, %r15
> >+      mov     %rdi, %r15
> >
> >       /* Set up the stack for verify_cpu() */
> >       leaq    __top_init_kernel_stack(%rip), %rsp

> This is an ABI.
>
> Nacked-by: H. Peter Anvin <hpa@zytor.com>

Peter, thanks for the response but if i ask you:
What is specific reason that this patch is wrong?
Why used RSI instead of RDI at the first place?

I broke my patch reference and it became stand alone each one, however this is one of three series. 
This makes the entry more ABI complaint, if there is no specific reason of objection.

Thanks

Re: [PATCH v1 1/3] Recieve boot_param from RDI Instead of RSI

[Brian Gerst]

On Mon, Jun 23, 2025 at 12:27 PM Khalid Ali <khaliidcaliy@gmail.com> wrote:
>
> > On June 23, 2025 1:15:23 AM PDT, Khalid Ali <khaliidcaliy@gmail.com> wrote:
> > >From: Khalid Ali <khaliidcaliy@gmail.com>
> > >
> > >Adjust the kernel entry point to recieve arguments from RDI instead of
> > >RSI.
> > >
> > >Signed-off-by: Khalid Ali <khaliidcaliy@gmail.com>
> > >---
> > > arch/x86/kernel/head_64.S | 4 ++--
> > > 1 file changed, 2 insertions(+), 2 deletions(-)
> > >
> > >diff --git a/arch/x86/kernel/head_64.S b/arch/x86/kernel/head_64.S
> > >index dfb5390e5c9a..d24fea15b6a6 100644
> > >--- a/arch/x86/kernel/head_64.S
> > >+++ b/arch/x86/kernel/head_64.S
> > >@@ -43,7 +43,7 @@ SYM_CODE_START_NOALIGN(startup_64)
> > >        * for us.  These identity mapped page tables map all of the
> > >        * kernel pages and possibly all of memory.
> > >        *
> > >-       * %RSI holds the physical address of the boot_params structure
> > >+       * %RDI holds the physical address of the boot_params structure
> > >        * provided by the bootloader. Preserve it in %R15 so C function calls
> > >        * will not clobber it.
> > >        *
> > >@@ -56,7 +56,7 @@ SYM_CODE_START_NOALIGN(startup_64)
> > >        * compiled to run at we first fixup the physical addresses in our page
> > >        * tables and then reload them.
> > >        */
> > >-      mov     %rsi, %r15
> > >+      mov     %rdi, %r15
> > >
> > >       /* Set up the stack for verify_cpu() */
> > >       leaq    __top_init_kernel_stack(%rip), %rsp
>
> > This is an ABI.
> >
> > Nacked-by: H. Peter Anvin <hpa@zytor.com>
>
> Peter, thanks for the response but if i ask you:
> What is specific reason that this patch is wrong?
> Why used RSI instead of RDI at the first place?

It is defined in Documentation/arch/x86/boot.rst

IIRC some hypervisors can boot the uncompressed image directly.


Brian Gerst