The secondary hart booting and relocation code are under .init section.
As a result, it will be freed once kernel booting is done. However,
ordered booting protocol and CPU hotplug always requires these sections
to be present to bringup harts after initial kernel boot.
Move the required sections to a different section and make sure that
they are in memory within first 2MB offset as trampoline page directory
only maps first 2MB.
Signed-off-by: Atish Patra <[email protected]>
---
arch/riscv/kernel/head.S | 73 +++++++++++++++++++--------------
arch/riscv/kernel/vmlinux.lds.S | 9 +++-
2 files changed, 50 insertions(+), 32 deletions(-)
diff --git a/arch/riscv/kernel/head.S b/arch/riscv/kernel/head.S
index a4242be66966..9d7f084a50cc 100644
--- a/arch/riscv/kernel/head.S
+++ b/arch/riscv/kernel/head.S
@@ -14,7 +14,7 @@
#include <asm/hwcap.h>
#include <asm/image.h>
-__INIT
+__HEAD
ENTRY(_start)
/*
* Image header expected by Linux boot-loaders. The image header data
@@ -44,9 +44,10 @@ ENTRY(_start)
.balign 4
.ascii RISCV_IMAGE_MAGIC2
.word 0
+END(_start)
-.global _start_kernel
-_start_kernel:
+ __INIT
+ENTRY(_start_kernel)
/* Mask all interrupts */
csrw CSR_IE, zero
csrw CSR_IP, zero
@@ -125,6 +126,37 @@ clear_bss_done:
call parse_dtb
tail start_kernel
+.Lsecondary_start:
+#ifdef CONFIG_SMP
+ /* Set trap vector to spin forever to help debug */
+ la a3, .Lsecondary_park
+ csrw CSR_TVEC, a3
+
+ slli a3, a0, LGREG
+ la a1, __cpu_up_stack_pointer
+ la a2, __cpu_up_task_pointer
+ add a1, a3, a1
+ add a2, a3, a2
+
+ /*
+ * This hart didn't win the lottery, so we wait for the winning hart to
+ * get far enough along the boot process that it should continue.
+ */
+.Lwait_for_cpu_up:
+ /* FIXME: We should WFI to save some energy here. */
+ REG_L sp, (a1)
+ REG_L tp, (a2)
+ beqz sp, .Lwait_for_cpu_up
+ beqz tp, .Lwait_for_cpu_up
+ fence
+
+ tail secondary_start_common
+#endif
+
+END(_start_kernel)
+
+.section ".noinit.text","ax",@progbits
+.align 2
#ifdef CONFIG_MMU
relocate:
/* Relocate return address */
@@ -177,41 +209,27 @@ relocate:
ret
#endif /* CONFIG_MMU */
-
-.Lsecondary_start:
#ifdef CONFIG_SMP
/* Set trap vector to spin forever to help debug */
la a3, .Lsecondary_park
csrw CSR_TVEC, a3
slli a3, a0, LGREG
- la a1, __cpu_up_stack_pointer
- la a2, __cpu_up_task_pointer
- add a1, a3, a1
- add a2, a3, a2
-
- /*
- * This hart didn't win the lottery, so we wait for the winning hart to
- * get far enough along the boot process that it should continue.
- */
-.Lwait_for_cpu_up:
- /* FIXME: We should WFI to save some energy here. */
- REG_L sp, (a1)
- REG_L tp, (a2)
- beqz sp, .Lwait_for_cpu_up
- beqz tp, .Lwait_for_cpu_up
- fence
+ .global secondary_start_common
+secondary_start_common:
#ifdef CONFIG_MMU
/* Enable virtual memory and relocate to virtual address */
la a0, swapper_pg_dir
call relocate
#endif
-
tail smp_callin
-#endif
+#endif /* CONFIG_SMP */
-END(_start)
+.Lsecondary_park:
+ /* We lack SMP support or have too many harts, so park this hart */
+ wfi
+ j .Lsecondary_park
#ifdef CONFIG_RISCV_M_MODE
ENTRY(reset_regs)
@@ -292,13 +310,6 @@ ENTRY(reset_regs)
END(reset_regs)
#endif /* CONFIG_RISCV_M_MODE */
-.section ".text", "ax",@progbits
-.align 2
-.Lsecondary_park:
- /* We lack SMP support or have too many harts, so park this hart */
- wfi
- j .Lsecondary_park
-
__PAGE_ALIGNED_BSS
/* Empty zero page */
.balign PAGE_SIZE
diff --git a/arch/riscv/kernel/vmlinux.lds.S b/arch/riscv/kernel/vmlinux.lds.S
index 12f42f96d46e..c8a88326df9e 100644
--- a/arch/riscv/kernel/vmlinux.lds.S
+++ b/arch/riscv/kernel/vmlinux.lds.S
@@ -10,6 +10,7 @@
#include <asm/cache.h>
#include <asm/thread_info.h>
+#include <linux/sizes.h>
OUTPUT_ARCH(riscv)
ENTRY(_start)
@@ -20,8 +21,14 @@ SECTIONS
/* Beginning of code and text segment */
. = LOAD_OFFSET;
_start = .;
- __init_begin = .;
HEAD_TEXT_SECTION
+ .noinit.text :
+ {
+ *(.noinit.text)
+ }
+ . = ALIGN(SZ_4K);
+
+ __init_begin = .;
INIT_TEXT_SECTION(PAGE_SIZE)
INIT_DATA_SECTION(16)
/* we have to discard exit text and such at runtime, not link time */
--
2.24.0
On Tue, Jan 28, 2020 at 7:58 AM Atish Patra <[email protected]> wrote:
>
> The secondary hart booting and relocation code are under .init section.
> As a result, it will be freed once kernel booting is done. However,
> ordered booting protocol and CPU hotplug always requires these sections
> to be present to bringup harts after initial kernel boot.
>
> Move the required sections to a different section and make sure that
> they are in memory within first 2MB offset as trampoline page directory
> only maps first 2MB.
>
> Signed-off-by: Atish Patra <[email protected]>
> ---
> arch/riscv/kernel/head.S | 73 +++++++++++++++++++--------------
> arch/riscv/kernel/vmlinux.lds.S | 9 +++-
> 2 files changed, 50 insertions(+), 32 deletions(-)
>
> diff --git a/arch/riscv/kernel/head.S b/arch/riscv/kernel/head.S
> index a4242be66966..9d7f084a50cc 100644
> --- a/arch/riscv/kernel/head.S
> +++ b/arch/riscv/kernel/head.S
> @@ -14,7 +14,7 @@
> #include <asm/hwcap.h>
> #include <asm/image.h>
>
> -__INIT
> +__HEAD
> ENTRY(_start)
> /*
> * Image header expected by Linux boot-loaders. The image header data
> @@ -44,9 +44,10 @@ ENTRY(_start)
> .balign 4
> .ascii RISCV_IMAGE_MAGIC2
> .word 0
> +END(_start)
>
> -.global _start_kernel
> -_start_kernel:
> + __INIT
> +ENTRY(_start_kernel)
> /* Mask all interrupts */
> csrw CSR_IE, zero
> csrw CSR_IP, zero
> @@ -125,6 +126,37 @@ clear_bss_done:
> call parse_dtb
> tail start_kernel
>
> +.Lsecondary_start:
> +#ifdef CONFIG_SMP
> + /* Set trap vector to spin forever to help debug */
> + la a3, .Lsecondary_park
> + csrw CSR_TVEC, a3
> +
> + slli a3, a0, LGREG
> + la a1, __cpu_up_stack_pointer
> + la a2, __cpu_up_task_pointer
> + add a1, a3, a1
> + add a2, a3, a2
> +
> + /*
> + * This hart didn't win the lottery, so we wait for the winning hart to
> + * get far enough along the boot process that it should continue.
> + */
> +.Lwait_for_cpu_up:
> + /* FIXME: We should WFI to save some energy here. */
> + REG_L sp, (a1)
> + REG_L tp, (a2)
> + beqz sp, .Lwait_for_cpu_up
> + beqz tp, .Lwait_for_cpu_up
> + fence
> +
> + tail secondary_start_common
> +#endif
> +
> +END(_start_kernel)
> +
> +.section ".noinit.text","ax",@progbits
> +.align 2
Try to use __HEAD here (if possible).
> #ifdef CONFIG_MMU
> relocate:
> /* Relocate return address */
> @@ -177,41 +209,27 @@ relocate:
>
> ret
> #endif /* CONFIG_MMU */
> -
> -.Lsecondary_start:
> #ifdef CONFIG_SMP
> /* Set trap vector to spin forever to help debug */
> la a3, .Lsecondary_park
> csrw CSR_TVEC, a3
>
> slli a3, a0, LGREG
> - la a1, __cpu_up_stack_pointer
> - la a2, __cpu_up_task_pointer
> - add a1, a3, a1
> - add a2, a3, a2
> -
> - /*
> - * This hart didn't win the lottery, so we wait for the winning hart to
> - * get far enough along the boot process that it should continue.
> - */
> -.Lwait_for_cpu_up:
> - /* FIXME: We should WFI to save some energy here. */
> - REG_L sp, (a1)
> - REG_L tp, (a2)
> - beqz sp, .Lwait_for_cpu_up
> - beqz tp, .Lwait_for_cpu_up
> - fence
> + .global secondary_start_common
> +secondary_start_common:
>
> #ifdef CONFIG_MMU
> /* Enable virtual memory and relocate to virtual address */
> la a0, swapper_pg_dir
> call relocate
> #endif
> -
> tail smp_callin
> -#endif
> +#endif /* CONFIG_SMP */
>
> -END(_start)
> +.Lsecondary_park:
> + /* We lack SMP support or have too many harts, so park this hart */
> + wfi
> + j .Lsecondary_park
>
> #ifdef CONFIG_RISCV_M_MODE
> ENTRY(reset_regs)
> @@ -292,13 +310,6 @@ ENTRY(reset_regs)
> END(reset_regs)
> #endif /* CONFIG_RISCV_M_MODE */
>
> -.section ".text", "ax",@progbits
> -.align 2
> -.Lsecondary_park:
> - /* We lack SMP support or have too many harts, so park this hart */
> - wfi
> - j .Lsecondary_park
> -
> __PAGE_ALIGNED_BSS
> /* Empty zero page */
> .balign PAGE_SIZE
> diff --git a/arch/riscv/kernel/vmlinux.lds.S b/arch/riscv/kernel/vmlinux.lds.S
> index 12f42f96d46e..c8a88326df9e 100644
> --- a/arch/riscv/kernel/vmlinux.lds.S
> +++ b/arch/riscv/kernel/vmlinux.lds.S
> @@ -10,6 +10,7 @@
> #include <asm/cache.h>
> #include <asm/thread_info.h>
>
> +#include <linux/sizes.h>
> OUTPUT_ARCH(riscv)
> ENTRY(_start)
>
> @@ -20,8 +21,14 @@ SECTIONS
> /* Beginning of code and text segment */
> . = LOAD_OFFSET;
> _start = .;
> - __init_begin = .;
> HEAD_TEXT_SECTION
> + .noinit.text :
> + {
> + *(.noinit.text)
> + }
Can we try using HEAD_TEXT_SECTION for SMP booting
related functions instead of new ".noinit.text" section ??
> + . = ALIGN(SZ_4K);
Change this to PAGE aligned:
. = ALIGN(PAGE_SIZE)
> +
> + __init_begin = .;
> INIT_TEXT_SECTION(PAGE_SIZE)
> INIT_DATA_SECTION(16)
> /* we have to discard exit text and such at runtime, not link time */
> --
> 2.24.0
>
Regards,
Anup
On Mon, Jan 27, 2020 at 8:38 PM Anup Patel <[email protected]> wrote:
>
> On Tue, Jan 28, 2020 at 7:58 AM Atish Patra <[email protected]> wrote:
> >
> > The secondary hart booting and relocation code are under .init section.
> > As a result, it will be freed once kernel booting is done. However,
> > ordered booting protocol and CPU hotplug always requires these sections
> > to be present to bringup harts after initial kernel boot.
> >
> > Move the required sections to a different section and make sure that
> > they are in memory within first 2MB offset as trampoline page directory
> > only maps first 2MB.
> >
> > Signed-off-by: Atish Patra <[email protected]>
> > ---
> > arch/riscv/kernel/head.S | 73 +++++++++++++++++++--------------
> > arch/riscv/kernel/vmlinux.lds.S | 9 +++-
> > 2 files changed, 50 insertions(+), 32 deletions(-)
> >
> > diff --git a/arch/riscv/kernel/head.S b/arch/riscv/kernel/head.S
> > index a4242be66966..9d7f084a50cc 100644
> > --- a/arch/riscv/kernel/head.S
> > +++ b/arch/riscv/kernel/head.S
> > @@ -14,7 +14,7 @@
> > #include <asm/hwcap.h>
> > #include <asm/image.h>
> >
> > -__INIT
> > +__HEAD
> > ENTRY(_start)
> > /*
> > * Image header expected by Linux boot-loaders. The image header data
> > @@ -44,9 +44,10 @@ ENTRY(_start)
> > .balign 4
> > .ascii RISCV_IMAGE_MAGIC2
> > .word 0
> > +END(_start)
> >
> > -.global _start_kernel
> > -_start_kernel:
> > + __INIT
> > +ENTRY(_start_kernel)
> > /* Mask all interrupts */
> > csrw CSR_IE, zero
> > csrw CSR_IP, zero
> > @@ -125,6 +126,37 @@ clear_bss_done:
> > call parse_dtb
> > tail start_kernel
> >
> > +.Lsecondary_start:
> > +#ifdef CONFIG_SMP
> > + /* Set trap vector to spin forever to help debug */
> > + la a3, .Lsecondary_park
> > + csrw CSR_TVEC, a3
> > +
> > + slli a3, a0, LGREG
> > + la a1, __cpu_up_stack_pointer
> > + la a2, __cpu_up_task_pointer
> > + add a1, a3, a1
> > + add a2, a3, a2
> > +
> > + /*
> > + * This hart didn't win the lottery, so we wait for the winning hart to
> > + * get far enough along the boot process that it should continue.
> > + */
> > +.Lwait_for_cpu_up:
> > + /* FIXME: We should WFI to save some energy here. */
> > + REG_L sp, (a1)
> > + REG_L tp, (a2)
> > + beqz sp, .Lwait_for_cpu_up
> > + beqz tp, .Lwait_for_cpu_up
> > + fence
> > +
> > + tail secondary_start_common
> > +#endif
> > +
> > +END(_start_kernel)
> > +
> > +.section ".noinit.text","ax",@progbits
> > +.align 2
>
> Try to use __HEAD here (if possible).
>
> > #ifdef CONFIG_MMU
> > relocate:
> > /* Relocate return address */
> > @@ -177,41 +209,27 @@ relocate:
> >
> > ret
> > #endif /* CONFIG_MMU */
> > -
> > -.Lsecondary_start:
> > #ifdef CONFIG_SMP
> > /* Set trap vector to spin forever to help debug */
> > la a3, .Lsecondary_park
> > csrw CSR_TVEC, a3
> >
> > slli a3, a0, LGREG
> > - la a1, __cpu_up_stack_pointer
> > - la a2, __cpu_up_task_pointer
> > - add a1, a3, a1
> > - add a2, a3, a2
> > -
> > - /*
> > - * This hart didn't win the lottery, so we wait for the winning hart to
> > - * get far enough along the boot process that it should continue.
> > - */
> > -.Lwait_for_cpu_up:
> > - /* FIXME: We should WFI to save some energy here. */
> > - REG_L sp, (a1)
> > - REG_L tp, (a2)
> > - beqz sp, .Lwait_for_cpu_up
> > - beqz tp, .Lwait_for_cpu_up
> > - fence
> > + .global secondary_start_common
> > +secondary_start_common:
> >
> > #ifdef CONFIG_MMU
> > /* Enable virtual memory and relocate to virtual address */
> > la a0, swapper_pg_dir
> > call relocate
> > #endif
> > -
> > tail smp_callin
> > -#endif
> > +#endif /* CONFIG_SMP */
> >
> > -END(_start)
> > +.Lsecondary_park:
> > + /* We lack SMP support or have too many harts, so park this hart */
> > + wfi
> > + j .Lsecondary_park
> >
> > #ifdef CONFIG_RISCV_M_MODE
> > ENTRY(reset_regs)
> > @@ -292,13 +310,6 @@ ENTRY(reset_regs)
> > END(reset_regs)
> > #endif /* CONFIG_RISCV_M_MODE */
> >
> > -.section ".text", "ax",@progbits
> > -.align 2
> > -.Lsecondary_park:
> > - /* We lack SMP support or have too many harts, so park this hart */
> > - wfi
> > - j .Lsecondary_park
> > -
> > __PAGE_ALIGNED_BSS
> > /* Empty zero page */
> > .balign PAGE_SIZE
> > diff --git a/arch/riscv/kernel/vmlinux.lds.S b/arch/riscv/kernel/vmlinux.lds.S
> > index 12f42f96d46e..c8a88326df9e 100644
> > --- a/arch/riscv/kernel/vmlinux.lds.S
> > +++ b/arch/riscv/kernel/vmlinux.lds.S
> > @@ -10,6 +10,7 @@
> > #include <asm/cache.h>
> > #include <asm/thread_info.h>
> >
> > +#include <linux/sizes.h>
> > OUTPUT_ARCH(riscv)
> > ENTRY(_start)
> >
> > @@ -20,8 +21,14 @@ SECTIONS
> > /* Beginning of code and text segment */
> > . = LOAD_OFFSET;
> > _start = .;
> > - __init_begin = .;
> > HEAD_TEXT_SECTION
> > + .noinit.text :
> > + {
> > + *(.noinit.text)
> > + }
>
> Can we try using HEAD_TEXT_SECTION for SMP booting
> related functions instead of new ".noinit.text" section ??
>
Yeah. That is much cleaner as well. I will update it in v2.
> > + . = ALIGN(SZ_4K);
>
> Change this to PAGE aligned:
> . = ALIGN(PAGE_SIZE)
>
Sure. Will do that.
> > +
> > + __init_begin = .;
> > INIT_TEXT_SECTION(PAGE_SIZE)
> > INIT_DATA_SECTION(16)
> > /* we have to discard exit text and such at runtime, not link time */
> > --
> > 2.24.0
> >
>
> Regards,
> Anup
>
--
Regards,
Atish