The Linux Kernel Mailing List
 help / color / mirror / Atom feed
* [PATCHv2 0/3] powerpc: kprobes insn page fixes
@ 2026-11-13 12:31 Ritesh Harjani (IBM)
  2026-11-13 12:31 ` [PATCHv2 1/3] powerpc/execmem: Don't leave kprobes insn pages W+X when CONFIG_MODULES=n Ritesh Harjani (IBM)
                   ` (2 more replies)
  0 siblings, 3 replies; 14+ messages in thread
From: Ritesh Harjani (IBM) @ 2026-11-13 12:31 UTC (permalink / raw)
  To: linuxppc-dev
  Cc: Madhavan Srinivasan, Venkat Rao Bagalkote, linux-kernel, chleroy,
	Michael Ellerman, Nicholas Piggin, Shrikanth Hegde, Hari Bathini,
	Ritesh Harjani (IBM)

Hi All,

A few fixes that came out of looking at a W+X warning at boot with
KPROBES=y and MODULES=n.

- Patch-1 fixes that warning.
- Patch-2 (reported by Sashiko) fixes an oops on book3s/32 when a kprobe has to
  single step an instruction out of line, the insn page was in an NX segment.
- Patch-3 makes ptdump show "X" for kernel mappings on e500, which was noticed
  while testing the other two.

Ritesh Harjani (IBM) (3):
  powerpc/execmem: Don't leave kprobes insn pages W+X when CONFIG_MODULES=n
  powerpc/execmem: Allocate kprobes insn pages from the modules area on book3s/32
  powerpc/ptdump: Show the exec flag of kernel mappings on e500

 arch/powerpc/mm/mem.c           | 12 +++++++++---
 arch/powerpc/mm/ptdump/shared.c |  6 +++---
 2 files changed, 12 insertions(+), 6 deletions(-)

--
2.39.5


^ permalink raw reply	[flat|nested] 14+ messages in thread

* [PATCHv2 1/3] powerpc/execmem: Don't leave kprobes insn pages W+X when CONFIG_MODULES=n
  2026-11-13 12:31 [PATCHv2 0/3] powerpc: kprobes insn page fixes Ritesh Harjani (IBM)
@ 2026-11-13 12:31 ` Ritesh Harjani (IBM)
  2026-11-17 14:08   ` Christophe Leroy (CS GROUP)
  2026-11-13 12:31 ` [PATCHv2 2/3] powerpc/execmem: Allocate kprobes insn pages from the modules area on book3s/32 Ritesh Harjani (IBM)
  2026-11-13 12:31 ` [PATCHv2 3/3] powerpc/ptdump: Show the exec flag of kernel mappings on e500 Ritesh Harjani (IBM)
  2 siblings, 1 reply; 14+ messages in thread
From: Ritesh Harjani (IBM) @ 2026-11-13 12:31 UTC (permalink / raw)
  To: linuxppc-dev
  Cc: Madhavan Srinivasan, Venkat Rao Bagalkote, linux-kernel, chleroy,
	Michael Ellerman, Nicholas Piggin, Shrikanth Hegde, Hari Bathini,
	Ritesh Harjani (IBM)

commit 2e83e2ce34ff ("arch, mm: promote DEBUG_WX to CHECK_WX"), made
CHECK_WX default STRICT_KERNEL_RWX.

Now on powerpc, when we boot with CONFIG_KPROBES=y and CONFIG_MODULES=n,
we can see the following warning splat at boot:

    powerpc/mm: Found insecure W+X mapping at address 000000002ea08992/0xc008000000000000
    WARNING: arch/powerpc/mm/ptdump/ptdump.c:194 at note_page+0x4dc/0x4fc, CPU#0: swapper/0/1
    NIP [c0000000000ac204] note_page+0x4dc/0x4fc
    LR [c0000000000ac200] note_page+0x4d8/0x4fc
    Call Trace:
     note_page+0x4d8/0x4fc (unreliable)
     ptdump_pte_entry+0x70/0x94
     walk_pte_range_inner+0x15c/0x260
     walk_pgd_range+0x6d8/0xd5c
     walk_page_range_debug+0xa4/0xf0
     ptdump_walk_pgd+0xec/0x304
     ptdump_check_wx+0xd8/0x154
     kernel_init+0x94/0x26c
     ret_from_kernel_user_thread+0x14/0x1c
    Checked W+X mappings: FAILED, 1 W+X pages found

This happens because CHECK_WX noted a writeable + executable mapping.

This happens because execmem_arch_setup(), checks for
struct_modules_rwx_enabled(), while kprobes depenency on modules was
earlier removed via commit:
    7582b7be16d0 ("kprobes: remove dependency on CONFIG_MODULES")
Since the insn slots are only written with patch_instruction(), which
does not need this mapping to be writable. This patch fixes it by
checking for strict_kernel_rwx_enabled() instead.

Fixes: 7582b7be16d0 ("kprobes: remove dependency on CONFIG_MODULES")
Signed-off-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
---
 arch/powerpc/mm/mem.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
index b617b69452cd..77b63b52d8f2 100644
--- a/arch/powerpc/mm/mem.c
+++ b/arch/powerpc/mm/mem.c
@@ -388,7 +388,7 @@ static void prealloc_execmem_pgtable(void) { }
 
 struct execmem_info __init *execmem_arch_setup(void)
 {
-	pgprot_t kprobes_prot = strict_module_rwx_enabled() ? PAGE_KERNEL_ROX : PAGE_KERNEL_EXEC;
+	pgprot_t kprobes_prot = strict_kernel_rwx_enabled() ? PAGE_KERNEL_ROX : PAGE_KERNEL_EXEC;
 	pgprot_t prot = strict_module_rwx_enabled() ? PAGE_KERNEL : PAGE_KERNEL_EXEC;
 	unsigned long fallback_start = 0, fallback_end = 0;
 	unsigned long start, end;
-- 
2.39.5


^ permalink raw reply related	[flat|nested] 14+ messages in thread

* [PATCHv2 2/3] powerpc/execmem: Allocate kprobes insn pages from the modules area on book3s/32
  2026-11-13 12:31 [PATCHv2 0/3] powerpc: kprobes insn page fixes Ritesh Harjani (IBM)
  2026-11-13 12:31 ` [PATCHv2 1/3] powerpc/execmem: Don't leave kprobes insn pages W+X when CONFIG_MODULES=n Ritesh Harjani (IBM)
@ 2026-11-13 12:31 ` Ritesh Harjani (IBM)
  2026-11-15  8:29   ` Mike Rapoport
  2026-11-13 12:31 ` [PATCHv2 3/3] powerpc/ptdump: Show the exec flag of kernel mappings on e500 Ritesh Harjani (IBM)
  2 siblings, 1 reply; 14+ messages in thread
From: Ritesh Harjani (IBM) @ 2026-11-13 12:31 UTC (permalink / raw)
  To: linuxppc-dev
  Cc: Madhavan Srinivasan, Venkat Rao Bagalkote, linux-kernel, chleroy,
	Michael Ellerman, Nicholas Piggin, Shrikanth Hegde, Hari Bathini,
	Ritesh Harjani (IBM)

On book3s/32, a kprobe on an instruction that can't be emulated and has
to be single stepped out of line, crashes the kernel. For e.g. with the
following cmds...
    ~> echo 9 > /proc/sys/kernel/printk
    ~> mount -t debugfs debugfs /sys/kernel/debug
    ~> echo 'p:wx filldir64+0xb8' > /sys/kernel/debug/tracing/kprobe_events
    ~> echo 1 > /sys/kernel/debug/tracing/events/kprobes/wx/enable
    ~> echo 1 > /sys/kernel/debug/tracing/events/kprobes/enable

...we can see the following crash:
    kernel tried to execute exec-protected page (f101a008) - exploit attempt? (uid: 0)
    BUG: Unable to handle kernel instruction fetch
    Faulting instruction address: 0xc02b919c
    Vector: 400 (Instruction Access) at [f19a9d60]
        pc: c02b919c: filldir64+0xb8/0x2dc
        lr: c02b9148: filldir64+0x64/0x2dc
        sp: f19a9e20
       msr: 10029032
      current = 0xc112e900
        pid   = 102, comm = ls
    enter ? for help
    [f19a9e20] c1a8d420 (unreliable)
    [f19a9e50] c02e4074 dcache_readdir+0x208/0x2a8
    [f19a9e90] c02b8c5c iterate_dir+0x88/0x1fc
    [f19a9ec0] c02b96e0 sys_getdents64+0x88/0x198
    [f19a9f10] c0014400 system_call_exception+0xf8/0x20c
    [f19a9f30] c00191ac ret_from_syscall+0x0/0x2c
    ---- Exception: c00 (System Call) at a7596164
    SP (af802310) is in userspace

This happens because on book3s/32 the exec permission is per segment
(256MB), not per page. The page tables have no hardware exec bit, only
the segment registers have a no-execute (N) bit. mmu_mark_initmem_nx()
sets N on all the kernel segments except the ones of the modules area.
So nothing in the vmalloc area can be executed, even when the page is
mapped with PAGE_KERNEL_ROX or PAGE_KERNEL_EXEC, because _PAGE_EXEC is
a software only bit there and the segment's N bit is what the MMU
enforces.

Before commit 1b750c2fbf82 ("powerpc: extend execmem_params for kprobes
allocations"), the kprobes insn pages came from module_alloc(). That
commit moved them to the whole of the vmalloc space, on the grounds
that kprobes use a breakpoint and don't need to be near the kernel
text. That is fine on platforms where the exec permission is per page,
as PAGE_KERNEL_ROX is then enough to make a vmalloc page executable.
It is not on book3s/32, where the modules area is not only close to
the kernel text but is also the only page-mapped area whose segment is
left executable. So the insn pages must keep coming from there.

Hence, use the modules area for EXECMEM_KPROBES on book3s/32.

Fixes: 1b750c2fbf82 ("powerpc: extend execmem_params for kprobes allocations")
Signed-off-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
---
 arch/powerpc/mm/mem.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
index 77b63b52d8f2..ae1bc7359353 100644
--- a/arch/powerpc/mm/mem.c
+++ b/arch/powerpc/mm/mem.c
@@ -391,6 +391,7 @@ struct execmem_info __init *execmem_arch_setup(void)
 	pgprot_t kprobes_prot = strict_kernel_rwx_enabled() ? PAGE_KERNEL_ROX : PAGE_KERNEL_EXEC;
 	pgprot_t prot = strict_module_rwx_enabled() ? PAGE_KERNEL : PAGE_KERNEL_EXEC;
 	unsigned long fallback_start = 0, fallback_end = 0;
+	unsigned long kprobes_start = VMALLOC_START, kprobes_end = VMALLOC_END;
 	unsigned long start, end;

 	/*
@@ -410,6 +411,11 @@ struct execmem_info __init *execmem_arch_setup(void)
 	}

 	end = MODULES_END;
+
+	if (IS_ENABLED(CONFIG_PPC_BOOK3S_32)) {
+		kprobes_start = MODULES_VADDR;
+		kprobes_end = MODULES_END;
+	}
 #else
 	start = VMALLOC_START;
 	end = VMALLOC_END;
@@ -428,8 +434,8 @@ struct execmem_info __init *execmem_arch_setup(void)
 				.fallback_end	= fallback_end,
 			},
 			[EXECMEM_KPROBES] = {
-				.start	= VMALLOC_START,
-				.end	= VMALLOC_END,
+				.start	= kprobes_start,
+				.end	= kprobes_end,
 				.pgprot	= kprobes_prot,
 				.alignment = 1,
 			},
--
2.39.5


^ permalink raw reply related	[flat|nested] 14+ messages in thread

* [PATCHv2 3/3] powerpc/ptdump: Show the exec flag of kernel mappings on e500
  2026-11-13 12:31 [PATCHv2 0/3] powerpc: kprobes insn page fixes Ritesh Harjani (IBM)
  2026-11-13 12:31 ` [PATCHv2 1/3] powerpc/execmem: Don't leave kprobes insn pages W+X when CONFIG_MODULES=n Ritesh Harjani (IBM)
  2026-11-13 12:31 ` [PATCHv2 2/3] powerpc/execmem: Allocate kprobes insn pages from the modules area on book3s/32 Ritesh Harjani (IBM)
@ 2026-11-13 12:31 ` Ritesh Harjani (IBM)
  2026-11-17 14:09   ` Christophe Leroy (CS GROUP)
  2 siblings, 1 reply; 14+ messages in thread
From: Ritesh Harjani (IBM) @ 2026-11-13 12:31 UTC (permalink / raw)
  To: linuxppc-dev
  Cc: Madhavan Srinivasan, Venkat Rao Bagalkote, linux-kernel, chleroy,
	Michael Ellerman, Nicholas Piggin, Shrikanth Hegde, Hari Bathini,
	Ritesh Harjani (IBM)

On e500, _PAGE_EXEC is made of two bits, _PAGE_BAP_SX and _PAGE_BAP_UX,
and a kernel mapping only has _PAGE_BAP_SX (supervisor bit). The flag
table requires all the bits of the mask to be set to print "X", so
kernel_page_tables never shows a kernel mapping as executable:

w/o this patch:
  0xf1015000-0xf1015fff  0x00000000         4K PTE   r          present  coherent                  accessed

Print "X" when any of the bits is set the same way it is done for
_PAGE_READ and _PAGE_WRITE:

with this patch:
  0xf1015000-0xf1015fff  0x00000000         4K PTE   r      X   present  coherent                  accessed

This only changes what is displayed. ptdump_check_wx() uses pte_exec()
which already tests for either of the bits.

Signed-off-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
---
 arch/powerpc/mm/ptdump/shared.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/arch/powerpc/mm/ptdump/shared.c b/arch/powerpc/mm/ptdump/shared.c
index edc69da19b85..e28a41a8b1ac 100644
--- a/arch/powerpc/mm/ptdump/shared.c
+++ b/arch/powerpc/mm/ptdump/shared.c
@@ -22,9 +22,9 @@ static const struct flag_info flag_array[] = {
 		.clear	= "w",
 	}, {
 		.mask	= _PAGE_EXEC,
-		.val	= _PAGE_EXEC,
-		.set	= " X ",
-		.clear	= "   ",
+		.val	= 0,
+		.set	= "   ",
+		.clear	= " X ",
 	}, {
 		.mask	= _PAGE_PRESENT,
 		.val	= _PAGE_PRESENT,
--
2.39.5


^ permalink raw reply related	[flat|nested] 14+ messages in thread

* Re: [PATCHv2 2/3] powerpc/execmem: Allocate kprobes insn pages from the modules area on book3s/32
  2026-11-13 12:31 ` [PATCHv2 2/3] powerpc/execmem: Allocate kprobes insn pages from the modules area on book3s/32 Ritesh Harjani (IBM)
@ 2026-11-15  8:29   ` Mike Rapoport
  2026-11-16 15:11     ` Christophe Leroy (CS GROUP)
  0 siblings, 1 reply; 14+ messages in thread
From: Mike Rapoport @ 2026-11-15  8:29 UTC (permalink / raw)
  To: Ritesh Harjani (IBM)
  Cc: linuxppc-dev, Madhavan Srinivasan, Venkat Rao Bagalkote,
	linux-kernel, chleroy, Michael Ellerman, Nicholas Piggin,
	Shrikanth Hegde, Hari Bathini

On Fri, Nov 13, 2026 at 06:01:37PM +0530, Ritesh Harjani (IBM) wrote:
> On book3s/32, a kprobe on an instruction that can't be emulated and has
> to be single stepped out of line, crashes the kernel. For e.g. with the
> following cmds...
>     ~> echo 9 > /proc/sys/kernel/printk
>     ~> mount -t debugfs debugfs /sys/kernel/debug
>     ~> echo 'p:wx filldir64+0xb8' > /sys/kernel/debug/tracing/kprobe_events
>     ~> echo 1 > /sys/kernel/debug/tracing/events/kprobes/wx/enable
>     ~> echo 1 > /sys/kernel/debug/tracing/events/kprobes/enable
> 
> ...we can see the following crash:
>     kernel tried to execute exec-protected page (f101a008) - exploit attempt? (uid: 0)
>     BUG: Unable to handle kernel instruction fetch
>     Faulting instruction address: 0xc02b919c
>     Vector: 400 (Instruction Access) at [f19a9d60]
>         pc: c02b919c: filldir64+0xb8/0x2dc
>         lr: c02b9148: filldir64+0x64/0x2dc
>         sp: f19a9e20
>        msr: 10029032
>       current = 0xc112e900
>         pid   = 102, comm = ls
>     enter ? for help
>     [f19a9e20] c1a8d420 (unreliable)
>     [f19a9e50] c02e4074 dcache_readdir+0x208/0x2a8
>     [f19a9e90] c02b8c5c iterate_dir+0x88/0x1fc
>     [f19a9ec0] c02b96e0 sys_getdents64+0x88/0x198
>     [f19a9f10] c0014400 system_call_exception+0xf8/0x20c
>     [f19a9f30] c00191ac ret_from_syscall+0x0/0x2c
>     ---- Exception: c00 (System Call) at a7596164
>     SP (af802310) is in userspace
> 
> This happens because on book3s/32 the exec permission is per segment
> (256MB), not per page. The page tables have no hardware exec bit, only
> the segment registers have a no-execute (N) bit. mmu_mark_initmem_nx()
> sets N on all the kernel segments except the ones of the modules area.
> So nothing in the vmalloc area can be executed, even when the page is
> mapped with PAGE_KERNEL_ROX or PAGE_KERNEL_EXEC, because _PAGE_EXEC is
> a software only bit there and the segment's N bit is what the MMU
> enforces.
> 
> Before commit 1b750c2fbf82 ("powerpc: extend execmem_params for kprobes
> allocations"), the kprobes insn pages came from module_alloc(). That
> commit moved them to the whole of the vmalloc space, on the grounds
> that kprobes use a breakpoint and don't need to be near the kernel
> text. That is fine on platforms where the exec permission is per page,
> as PAGE_KERNEL_ROX is then enough to make a vmalloc page executable.
> It is not on book3s/32, where the modules area is not only close to
> the kernel text but is also the only page-mapped area whose segment is
> left executable. So the insn pages must keep coming from there.
> 
> Hence, use the modules area for EXECMEM_KPROBES on book3s/32.
 
Right, I missed that on book3s/32 does vmalloc is not executable and its
protection can't be changed.

Could it be that 8xx also need a special treatment for kprobes?

> Fixes: 1b750c2fbf82 ("powerpc: extend execmem_params for kprobes allocations")
> Signed-off-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>

Reviewed-by: Mike Rapoport (Microsoft) <rppt@kernel.org>

> ---
>  arch/powerpc/mm/mem.c | 10 ++++++++--
>  1 file changed, 8 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
> index 77b63b52d8f2..ae1bc7359353 100644
> --- a/arch/powerpc/mm/mem.c
> +++ b/arch/powerpc/mm/mem.c
> @@ -391,6 +391,7 @@ struct execmem_info __init *execmem_arch_setup(void)
>  	pgprot_t kprobes_prot = strict_kernel_rwx_enabled() ? PAGE_KERNEL_ROX : PAGE_KERNEL_EXEC;
>  	pgprot_t prot = strict_module_rwx_enabled() ? PAGE_KERNEL : PAGE_KERNEL_EXEC;
>  	unsigned long fallback_start = 0, fallback_end = 0;
> +	unsigned long kprobes_start = VMALLOC_START, kprobes_end = VMALLOC_END;
>  	unsigned long start, end;
> 
>  	/*
> @@ -410,6 +411,11 @@ struct execmem_info __init *execmem_arch_setup(void)
>  	}
> 
>  	end = MODULES_END;
> +
> +	if (IS_ENABLED(CONFIG_PPC_BOOK3S_32)) {
> +		kprobes_start = MODULES_VADDR;
> +		kprobes_end = MODULES_END;
> +	}
>  #else
>  	start = VMALLOC_START;
>  	end = VMALLOC_END;
> @@ -428,8 +434,8 @@ struct execmem_info __init *execmem_arch_setup(void)
>  				.fallback_end	= fallback_end,
>  			},
>  			[EXECMEM_KPROBES] = {
> -				.start	= VMALLOC_START,
> -				.end	= VMALLOC_END,
> +				.start	= kprobes_start,
> +				.end	= kprobes_end,
>  				.pgprot	= kprobes_prot,
>  				.alignment = 1,
>  			},
> --
> 2.39.5
> 
> 

-- 
Sincerely yours,
Mike.

^ permalink raw reply	[flat|nested] 14+ messages in thread

* Re: [PATCHv2 2/3] powerpc/execmem: Allocate kprobes insn pages from the modules area on book3s/32
  2026-11-15  8:29   ` Mike Rapoport
@ 2026-11-16 15:11     ` Christophe Leroy (CS GROUP)
  2026-11-16 15:57       ` Mike Rapoport
  0 siblings, 1 reply; 14+ messages in thread
From: Christophe Leroy (CS GROUP) @ 2026-11-16 15:11 UTC (permalink / raw)
  To: Mike Rapoport, Ritesh Harjani (IBM)
  Cc: linuxppc-dev, Madhavan Srinivasan, Venkat Rao Bagalkote,
	linux-kernel, Michael Ellerman, Nicholas Piggin, Shrikanth Hegde,
	Hari Bathini

Hi Mike,

Le 15/11/2026 à 09:29, Mike Rapoport a écrit :
> On Fri, Nov 13, 2026 at 06:01:37PM +0530, Ritesh Harjani (IBM) wrote:
>> On book3s/32, a kprobe on an instruction that can't be emulated and has
>> to be single stepped out of line, crashes the kernel. For e.g. with the
>> following cmds...
>>      ~> echo 9 > /proc/sys/kernel/printk
>>      ~> mount -t debugfs debugfs /sys/kernel/debug
>>      ~> echo 'p:wx filldir64+0xb8' > /sys/kernel/debug/tracing/kprobe_events
>>      ~> echo 1 > /sys/kernel/debug/tracing/events/kprobes/wx/enable
>>      ~> echo 1 > /sys/kernel/debug/tracing/events/kprobes/enable
>>
>> ...we can see the following crash:
>>      kernel tried to execute exec-protected page (f101a008) - exploit attempt? (uid: 0)
>>      BUG: Unable to handle kernel instruction fetch
>>      Faulting instruction address: 0xc02b919c
>>      Vector: 400 (Instruction Access) at [f19a9d60]
>>          pc: c02b919c: filldir64+0xb8/0x2dc
>>          lr: c02b9148: filldir64+0x64/0x2dc
>>          sp: f19a9e20
>>         msr: 10029032
>>        current = 0xc112e900
>>          pid   = 102, comm = ls
>>      enter ? for help
>>      [f19a9e20] c1a8d420 (unreliable)
>>      [f19a9e50] c02e4074 dcache_readdir+0x208/0x2a8
>>      [f19a9e90] c02b8c5c iterate_dir+0x88/0x1fc
>>      [f19a9ec0] c02b96e0 sys_getdents64+0x88/0x198
>>      [f19a9f10] c0014400 system_call_exception+0xf8/0x20c
>>      [f19a9f30] c00191ac ret_from_syscall+0x0/0x2c
>>      ---- Exception: c00 (System Call) at a7596164
>>      SP (af802310) is in userspace
>>
>> This happens because on book3s/32 the exec permission is per segment
>> (256MB), not per page. The page tables have no hardware exec bit, only
>> the segment registers have a no-execute (N) bit. mmu_mark_initmem_nx()
>> sets N on all the kernel segments except the ones of the modules area.
>> So nothing in the vmalloc area can be executed, even when the page is
>> mapped with PAGE_KERNEL_ROX or PAGE_KERNEL_EXEC, because _PAGE_EXEC is
>> a software only bit there and the segment's N bit is what the MMU
>> enforces.
>>
>> Before commit 1b750c2fbf82 ("powerpc: extend execmem_params for kprobes
>> allocations"), the kprobes insn pages came from module_alloc(). That
>> commit moved them to the whole of the vmalloc space, on the grounds
>> that kprobes use a breakpoint and don't need to be near the kernel
>> text. That is fine on platforms where the exec permission is per page,
>> as PAGE_KERNEL_ROX is then enough to make a vmalloc page executable.
>> It is not on book3s/32, where the modules area is not only close to
>> the kernel text but is also the only page-mapped area whose segment is
>> left executable. So the insn pages must keep coming from there.
>>
>> Hence, use the modules area for EXECMEM_KPROBES on book3s/32.
>   
> Right, I missed that on book3s/32 does vmalloc is not executable and its
> protection can't be changed.
> 
> Could it be that 8xx also need a special treatment for kprobes?

Yes indeed, not for the same reasons but because 8xx ITLB-miss handler 
doesn't expect any instruction TLB miss on kernel addresses other than 
the ones from the module area:

Commit 4afb0be7ccda ("powerpc/8xx: Avoid testing for kernel space in 
ITLB Miss.")
Commit 16a71c045186 ("powerpc/8xx: Preallocate execmem page tables")
Commit 33c527522f39 ("powerpc/8xx: Inconditionally use task PGDIR in 
ITLB misses")

So at the end as MODULES_VADDR/MODULES_END only exist on book3s/32 and 
8xx I think when they exist they should be used as follows:

diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
index b617b69452cd..856f4609c488 100644
--- a/arch/powerpc/mm/mem.c
+++ b/arch/powerpc/mm/mem.c
@@ -428,8 +428,13 @@ struct execmem_info __init *execmem_arch_setup(void)
  				.fallback_end	= fallback_end,
  			},
  			[EXECMEM_KPROBES] = {
+#ifdef MODULES_VADDR
+				.start	= MODULES_VADDR,
+				.end	= MODULES_END,
+#else
  				.start	= VMALLOC_START,
  				.end	= VMALLOC_END,
+#endif
  				.pgprot	= kprobes_prot,
  				.alignment = 1,
  			},


> 
>> Fixes: 1b750c2fbf82 ("powerpc: extend execmem_params for kprobes allocations")
>> Signed-off-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
> 
> Reviewed-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
> 
>> ---
>>   arch/powerpc/mm/mem.c | 10 ++++++++--
>>   1 file changed, 8 insertions(+), 2 deletions(-)
>>
>> diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
>> index 77b63b52d8f2..ae1bc7359353 100644
>> --- a/arch/powerpc/mm/mem.c
>> +++ b/arch/powerpc/mm/mem.c
>> @@ -391,6 +391,7 @@ struct execmem_info __init *execmem_arch_setup(void)
>>   	pgprot_t kprobes_prot = strict_kernel_rwx_enabled() ? PAGE_KERNEL_ROX : PAGE_KERNEL_EXEC;
>>   	pgprot_t prot = strict_module_rwx_enabled() ? PAGE_KERNEL : PAGE_KERNEL_EXEC;
>>   	unsigned long fallback_start = 0, fallback_end = 0;
>> +	unsigned long kprobes_start = VMALLOC_START, kprobes_end = VMALLOC_END;
>>   	unsigned long start, end;
>>
>>   	/*
>> @@ -410,6 +411,11 @@ struct execmem_info __init *execmem_arch_setup(void)
>>   	}
>>
>>   	end = MODULES_END;
>> +
>> +	if (IS_ENABLED(CONFIG_PPC_BOOK3S_32)) {
>> +		kprobes_start = MODULES_VADDR;
>> +		kprobes_end = MODULES_END;
>> +	}
>>   #else
>>   	start = VMALLOC_START;
>>   	end = VMALLOC_END;
>> @@ -428,8 +434,8 @@ struct execmem_info __init *execmem_arch_setup(void)
>>   				.fallback_end	= fallback_end,
>>   			},
>>   			[EXECMEM_KPROBES] = {
>> -				.start	= VMALLOC_START,
>> -				.end	= VMALLOC_END,
>> +				.start	= kprobes_start,
>> +				.end	= kprobes_end,
>>   				.pgprot	= kprobes_prot,
>>   				.alignment = 1,
>>   			},
>> --
>> 2.39.5
>>
>>
> 


^ permalink raw reply related	[flat|nested] 14+ messages in thread

* Re: [PATCHv2 2/3] powerpc/execmem: Allocate kprobes insn pages from the modules area on book3s/32
  2026-11-16 15:11     ` Christophe Leroy (CS GROUP)
@ 2026-11-16 15:57       ` Mike Rapoport
  2026-11-17 12:33         ` Ritesh Harjani
  0 siblings, 1 reply; 14+ messages in thread
From: Mike Rapoport @ 2026-11-16 15:57 UTC (permalink / raw)
  To: Christophe Leroy (CS GROUP)
  Cc: Ritesh Harjani (IBM), linuxppc-dev, Madhavan Srinivasan,
	Venkat Rao Bagalkote, linux-kernel, Michael Ellerman,
	Nicholas Piggin, Shrikanth Hegde, Hari Bathini

On Mon, Nov 16, 2026 at 04:11:07PM +0100, Christophe Leroy (CS GROUP) wrote:
> Hi Mike,
> 
> Le 15/11/2026 à 09:29, Mike Rapoport a écrit :
> > On Fri, Nov 13, 2026 at 06:01:37PM +0530, Ritesh Harjani (IBM) wrote:
> > > On book3s/32, a kprobe on an instruction that can't be emulated and has
> > > to be single stepped out of line, crashes the kernel. For e.g. with the
> > > following cmds...
> > >      ~> echo 9 > /proc/sys/kernel/printk
> > >      ~> mount -t debugfs debugfs /sys/kernel/debug
> > >      ~> echo 'p:wx filldir64+0xb8' > /sys/kernel/debug/tracing/kprobe_events
> > >      ~> echo 1 > /sys/kernel/debug/tracing/events/kprobes/wx/enable
> > >      ~> echo 1 > /sys/kernel/debug/tracing/events/kprobes/enable
> > > 
> > > ...we can see the following crash:
> > >      kernel tried to execute exec-protected page (f101a008) - exploit attempt? (uid: 0)
> > >      BUG: Unable to handle kernel instruction fetch
> > >      Faulting instruction address: 0xc02b919c
> > >      Vector: 400 (Instruction Access) at [f19a9d60]
> > >          pc: c02b919c: filldir64+0xb8/0x2dc
> > >          lr: c02b9148: filldir64+0x64/0x2dc
> > >          sp: f19a9e20
> > >         msr: 10029032
> > >        current = 0xc112e900
> > >          pid   = 102, comm = ls
> > >      enter ? for help
> > >      [f19a9e20] c1a8d420 (unreliable)
> > >      [f19a9e50] c02e4074 dcache_readdir+0x208/0x2a8
> > >      [f19a9e90] c02b8c5c iterate_dir+0x88/0x1fc
> > >      [f19a9ec0] c02b96e0 sys_getdents64+0x88/0x198
> > >      [f19a9f10] c0014400 system_call_exception+0xf8/0x20c
> > >      [f19a9f30] c00191ac ret_from_syscall+0x0/0x2c
> > >      ---- Exception: c00 (System Call) at a7596164
> > >      SP (af802310) is in userspace
> > > 
> > > This happens because on book3s/32 the exec permission is per segment
> > > (256MB), not per page. The page tables have no hardware exec bit, only
> > > the segment registers have a no-execute (N) bit. mmu_mark_initmem_nx()
> > > sets N on all the kernel segments except the ones of the modules area.
> > > So nothing in the vmalloc area can be executed, even when the page is
> > > mapped with PAGE_KERNEL_ROX or PAGE_KERNEL_EXEC, because _PAGE_EXEC is
> > > a software only bit there and the segment's N bit is what the MMU
> > > enforces.
> > > 
> > > Before commit 1b750c2fbf82 ("powerpc: extend execmem_params for kprobes
> > > allocations"), the kprobes insn pages came from module_alloc(). That
> > > commit moved them to the whole of the vmalloc space, on the grounds
> > > that kprobes use a breakpoint and don't need to be near the kernel
> > > text. That is fine on platforms where the exec permission is per page,
> > > as PAGE_KERNEL_ROX is then enough to make a vmalloc page executable.
> > > It is not on book3s/32, where the modules area is not only close to
> > > the kernel text but is also the only page-mapped area whose segment is
> > > left executable. So the insn pages must keep coming from there.
> > > 
> > > Hence, use the modules area for EXECMEM_KPROBES on book3s/32.
> > Right, I missed that on book3s/32 does vmalloc is not executable and its
> > protection can't be changed.
> > 
> > Could it be that 8xx also need a special treatment for kprobes?
> 
> Yes indeed, not for the same reasons but because 8xx ITLB-miss handler
> doesn't expect any instruction TLB miss on kernel addresses other than the
> ones from the module area:
> 
> Commit 4afb0be7ccda ("powerpc/8xx: Avoid testing for kernel space in ITLB
> Miss.")
> Commit 16a71c045186 ("powerpc/8xx: Preallocate execmem page tables")
> Commit 33c527522f39 ("powerpc/8xx: Inconditionally use task PGDIR in ITLB
> misses")
> 
> So at the end as MODULES_VADDR/MODULES_END only exist on book3s/32 and 8xx I
> think when they exist they should be used as follows:
> 
> diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
> index b617b69452cd..856f4609c488 100644
> --- a/arch/powerpc/mm/mem.c
> +++ b/arch/powerpc/mm/mem.c
> @@ -428,8 +428,13 @@ struct execmem_info __init *execmem_arch_setup(void)
>  				.fallback_end	= fallback_end,
>  			},
>  			[EXECMEM_KPROBES] = {
> +#ifdef MODULES_VADDR
> +				.start	= MODULES_VADDR,
> +				.end	= MODULES_END,
> +#else
>  				.start	= VMALLOC_START,
>  				.end	= VMALLOC_END,
> +#endif

We already have #ifdef MODULES_VADDR a few lines above, I think the
assignment of kprobes_start and kprobes_end could go there instead of if
(IS_ENABLED(CONFIG_PPC_BOOK3S_32)).

>  				.pgprot	= kprobes_prot,
>  				.alignment = 1,
>  			},
> 
> 
> > 
> > > Fixes: 1b750c2fbf82 ("powerpc: extend execmem_params for kprobes allocations")
> > > Signed-off-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
> > 
> > Reviewed-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
> > 
> > > ---
> > >   arch/powerpc/mm/mem.c | 10 ++++++++--
> > >   1 file changed, 8 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
> > > index 77b63b52d8f2..ae1bc7359353 100644
> > > --- a/arch/powerpc/mm/mem.c
> > > +++ b/arch/powerpc/mm/mem.c
> > > @@ -391,6 +391,7 @@ struct execmem_info __init *execmem_arch_setup(void)
> > >   	pgprot_t kprobes_prot = strict_kernel_rwx_enabled() ? PAGE_KERNEL_ROX : PAGE_KERNEL_EXEC;
> > >   	pgprot_t prot = strict_module_rwx_enabled() ? PAGE_KERNEL : PAGE_KERNEL_EXEC;
> > >   	unsigned long fallback_start = 0, fallback_end = 0;
> > > +	unsigned long kprobes_start = VMALLOC_START, kprobes_end = VMALLOC_END;
> > >   	unsigned long start, end;
> > > 
> > >   	/*
> > > @@ -410,6 +411,11 @@ struct execmem_info __init *execmem_arch_setup(void)
> > >   	}
> > > 
> > >   	end = MODULES_END;
> > > +
> > > +	if (IS_ENABLED(CONFIG_PPC_BOOK3S_32)) {
> > > +		kprobes_start = MODULES_VADDR;
> > > +		kprobes_end = MODULES_END;
> > > +	}
> > >   #else
> > >   	start = VMALLOC_START;
> > >   	end = VMALLOC_END;
> > > @@ -428,8 +434,8 @@ struct execmem_info __init *execmem_arch_setup(void)
> > >   				.fallback_end	= fallback_end,
> > >   			},
> > >   			[EXECMEM_KPROBES] = {
> > > -				.start	= VMALLOC_START,
> > > -				.end	= VMALLOC_END,
> > > +				.start	= kprobes_start,
> > > +				.end	= kprobes_end,
> > >   				.pgprot	= kprobes_prot,
> > >   				.alignment = 1,
> > >   			},
> > > --
> > > 2.39.5
> > > 
> > > 
> > 
> 

-- 
Sincerely yours,
Mike.

^ permalink raw reply	[flat|nested] 14+ messages in thread

* Re: [PATCHv2 2/3] powerpc/execmem: Allocate kprobes insn pages from the modules area on book3s/32
  2026-11-16 15:57       ` Mike Rapoport
@ 2026-11-17 12:33         ` Ritesh Harjani
  2026-11-17 14:04           ` Christophe Leroy (CS GROUP)
  0 siblings, 1 reply; 14+ messages in thread
From: Ritesh Harjani @ 2026-11-17 12:33 UTC (permalink / raw)
  To: Mike Rapoport, Christophe Leroy (CS GROUP)
  Cc: linuxppc-dev, Madhavan Srinivasan, Venkat Rao Bagalkote,
	linux-kernel, Michael Ellerman, Nicholas Piggin, Shrikanth Hegde,
	Hari Bathini

Mike Rapoport <rppt@kernel.org> writes:

> On Mon, Nov 16, 2026 at 04:11:07PM +0100, Christophe Leroy (CS GROUP) wrote:
>> Hi Mike,
>> 
>> Le 15/11/2026 à 09:29, Mike Rapoport a écrit :
>> > On Fri, Nov 13, 2026 at 06:01:37PM +0530, Ritesh Harjani (IBM) wrote:
>> > > On book3s/32, a kprobe on an instruction that can't be emulated and has
>> > > to be single stepped out of line, crashes the kernel. For e.g. with the
>> > > following cmds...
>> > >      ~> echo 9 > /proc/sys/kernel/printk
>> > >      ~> mount -t debugfs debugfs /sys/kernel/debug
>> > >      ~> echo 'p:wx filldir64+0xb8' > /sys/kernel/debug/tracing/kprobe_events
>> > >      ~> echo 1 > /sys/kernel/debug/tracing/events/kprobes/wx/enable
>> > >      ~> echo 1 > /sys/kernel/debug/tracing/events/kprobes/enable
>> > > 
>> > > ...we can see the following crash:
>> > >      kernel tried to execute exec-protected page (f101a008) - exploit attempt? (uid: 0)
>> > >      BUG: Unable to handle kernel instruction fetch
>> > >      Faulting instruction address: 0xc02b919c
>> > >      Vector: 400 (Instruction Access) at [f19a9d60]
>> > >          pc: c02b919c: filldir64+0xb8/0x2dc
>> > >          lr: c02b9148: filldir64+0x64/0x2dc
>> > >          sp: f19a9e20
>> > >         msr: 10029032
>> > >        current = 0xc112e900
>> > >          pid   = 102, comm = ls
>> > >      enter ? for help
>> > >      [f19a9e20] c1a8d420 (unreliable)
>> > >      [f19a9e50] c02e4074 dcache_readdir+0x208/0x2a8
>> > >      [f19a9e90] c02b8c5c iterate_dir+0x88/0x1fc
>> > >      [f19a9ec0] c02b96e0 sys_getdents64+0x88/0x198
>> > >      [f19a9f10] c0014400 system_call_exception+0xf8/0x20c
>> > >      [f19a9f30] c00191ac ret_from_syscall+0x0/0x2c
>> > >      ---- Exception: c00 (System Call) at a7596164
>> > >      SP (af802310) is in userspace
>> > > 
>> > > This happens because on book3s/32 the exec permission is per segment
>> > > (256MB), not per page. The page tables have no hardware exec bit, only
>> > > the segment registers have a no-execute (N) bit. mmu_mark_initmem_nx()
>> > > sets N on all the kernel segments except the ones of the modules area.
>> > > So nothing in the vmalloc area can be executed, even when the page is
>> > > mapped with PAGE_KERNEL_ROX or PAGE_KERNEL_EXEC, because _PAGE_EXEC is
>> > > a software only bit there and the segment's N bit is what the MMU
>> > > enforces.
>> > > 
>> > > Before commit 1b750c2fbf82 ("powerpc: extend execmem_params for kprobes
>> > > allocations"), the kprobes insn pages came from module_alloc(). That
>> > > commit moved them to the whole of the vmalloc space, on the grounds
>> > > that kprobes use a breakpoint and don't need to be near the kernel
>> > > text. That is fine on platforms where the exec permission is per page,
>> > > as PAGE_KERNEL_ROX is then enough to make a vmalloc page executable.
>> > > It is not on book3s/32, where the modules area is not only close to
>> > > the kernel text but is also the only page-mapped area whose segment is
>> > > left executable. So the insn pages must keep coming from there.
>> > > 
>> > > Hence, use the modules area for EXECMEM_KPROBES on book3s/32.
>> > Right, I missed that on book3s/32 does vmalloc is not executable and its
>> > protection can't be changed.
>> > 
>> > Could it be that 8xx also need a special treatment for kprobes?
>> 
>> Yes indeed, not for the same reasons but because 8xx ITLB-miss handler
>> doesn't expect any instruction TLB miss on kernel addresses other than the
>> ones from the module area:
>> 
>> Commit 4afb0be7ccda ("powerpc/8xx: Avoid testing for kernel space in ITLB
>> Miss.")
>> Commit 16a71c045186 ("powerpc/8xx: Preallocate execmem page tables")
>> Commit 33c527522f39 ("powerpc/8xx: Inconditionally use task PGDIR in ITLB
>> misses")
>> 

Right. I looked at the above series which added all of this support [1].
So looks like 603 might also have this related problem (other than the
per segment N bit issue), since it's iTLB miss handler only walks the
task's PGDIR, unlike it's dTLB handler which falls back to reloading the
missing kernel PGD entry from swapper_pg_dir. So what I meant to say is
that, similar to 8xx, a new PGD entry created in the kernel's vmalloc
area for an insn page does not get copied over after the task is
created (/PGDIR allocated), on 603 as well.
But since MODLUES_VADDR is defined for both BOOK3S/32 and 8xx, so I guess it
gets covered by the same change anyways.

[1]: https://lore.kernel.org/linuxppc-dev/cover.1724173828.git.christophe.leroy@csgroup.eu/

>> So at the end as MODULES_VADDR/MODULES_END only exist on book3s/32 and 8xx I
>> think when they exist they should be used as follows:
>> 
>> diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
>> index b617b69452cd..856f4609c488 100644
>> --- a/arch/powerpc/mm/mem.c
>> +++ b/arch/powerpc/mm/mem.c
>> @@ -428,8 +428,13 @@ struct execmem_info __init *execmem_arch_setup(void)
>>  				.fallback_end	= fallback_end,
>>  			},
>>  			[EXECMEM_KPROBES] = {
>> +#ifdef MODULES_VADDR
>> +				.start	= MODULES_VADDR,
>> +				.end	= MODULES_END,
>> +#else
>>  				.start	= VMALLOC_START,
>>  				.end	= VMALLOC_END,
>> +#endif
>
> We already have #ifdef MODULES_VADDR a few lines above, I think the
> assignment of kprobes_start and kprobes_end could go there instead of if
> (IS_ENABLED(CONFIG_PPC_BOOK3S_32)).
>

Right. So let me use the following diff then? 

Please let me know if this diff looks good to you? If yes, then I can
send a v3.

Also - it will be good to know if patch-1 and patch-3 looks ok too,
before I spin a next revision?

Thanks Christophe & Mike, for reviewing and pointing this out!


diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
index 77b63b52d8f2..547120a30bee 100644
--- a/arch/powerpc/mm/mem.c
+++ b/arch/powerpc/mm/mem.c
@@ -391,6 +391,7 @@ struct execmem_info __init *execmem_arch_setup(void)
        pgprot_t kprobes_prot = strict_kernel_rwx_enabled() ? PAGE_KERNEL_ROX : PAGE_KERNEL_EXEC;
        pgprot_t prot = strict_module_rwx_enabled() ? PAGE_KERNEL : PAGE_KERNEL_EXEC;
        unsigned long fallback_start = 0, fallback_end = 0;
+       unsigned long kprobes_start = VMALLOC_START, kprobes_end = VMALLOC_END;
        unsigned long start, end;

        /*
@@ -410,6 +411,9 @@ struct execmem_info __init *execmem_arch_setup(void)
        }

        end = MODULES_END;
+
+       kprobes_start = MODULES_VADDR;
+       kprobes_end = MODULES_END;
 #else
        start = VMALLOC_START;
        end = VMALLOC_END;
@@ -428,8 +432,8 @@ struct execmem_info __init *execmem_arch_setup(void)
                                .fallback_end   = fallback_end,
                        },
                        [EXECMEM_KPROBES] = {
-                               .start  = VMALLOC_START,
-                               .end    = VMALLOC_END,
+                               .start  = kprobes_start,
+                               .end    = kprobes_end,
                                .pgprot = kprobes_prot,
                                .alignment = 1,
                        },



-ritesh


>>  				.pgprot	= kprobes_prot,
>>  				.alignment = 1,
>>  			},
>> 
>> 
>> > 
>> > > Fixes: 1b750c2fbf82 ("powerpc: extend execmem_params for kprobes allocations")
>> > > Signed-off-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
>> > 
>> > Reviewed-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
>> > 
>> > > ---
>> > >   arch/powerpc/mm/mem.c | 10 ++++++++--
>> > >   1 file changed, 8 insertions(+), 2 deletions(-)
>> > > 
>> > > diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
>> > > index 77b63b52d8f2..ae1bc7359353 100644
>> > > --- a/arch/powerpc/mm/mem.c
>> > > +++ b/arch/powerpc/mm/mem.c
>> > > @@ -391,6 +391,7 @@ struct execmem_info __init *execmem_arch_setup(void)
>> > >   	pgprot_t kprobes_prot = strict_kernel_rwx_enabled() ? PAGE_KERNEL_ROX : PAGE_KERNEL_EXEC;
>> > >   	pgprot_t prot = strict_module_rwx_enabled() ? PAGE_KERNEL : PAGE_KERNEL_EXEC;
>> > >   	unsigned long fallback_start = 0, fallback_end = 0;
>> > > +	unsigned long kprobes_start = VMALLOC_START, kprobes_end = VMALLOC_END;
>> > >   	unsigned long start, end;
>> > > 
>> > >   	/*
>> > > @@ -410,6 +411,11 @@ struct execmem_info __init *execmem_arch_setup(void)
>> > >   	}
>> > > 
>> > >   	end = MODULES_END;
>> > > +
>> > > +	if (IS_ENABLED(CONFIG_PPC_BOOK3S_32)) {
>> > > +		kprobes_start = MODULES_VADDR;
>> > > +		kprobes_end = MODULES_END;
>> > > +	}
>> > >   #else
>> > >   	start = VMALLOC_START;
>> > >   	end = VMALLOC_END;
>> > > @@ -428,8 +434,8 @@ struct execmem_info __init *execmem_arch_setup(void)
>> > >   				.fallback_end	= fallback_end,
>> > >   			},
>> > >   			[EXECMEM_KPROBES] = {
>> > > -				.start	= VMALLOC_START,
>> > > -				.end	= VMALLOC_END,
>> > > +				.start	= kprobes_start,
>> > > +				.end	= kprobes_end,
>> > >   				.pgprot	= kprobes_prot,
>> > >   				.alignment = 1,
>> > >   			},
>> > > --
>> > > 2.39.5
>> > > 
>> > > 
>> > 
>> 
>
> -- 
> Sincerely yours,
> Mike.

^ permalink raw reply related	[flat|nested] 14+ messages in thread

* Re: [PATCHv2 2/3] powerpc/execmem: Allocate kprobes insn pages from the modules area on book3s/32
  2026-11-17 12:33         ` Ritesh Harjani
@ 2026-11-17 14:04           ` Christophe Leroy (CS GROUP)
  2026-11-17 14:15             ` Mike Rapoport
  0 siblings, 1 reply; 14+ messages in thread
From: Christophe Leroy (CS GROUP) @ 2026-11-17 14:04 UTC (permalink / raw)
  To: Ritesh Harjani (IBM), Mike Rapoport
  Cc: linuxppc-dev, Madhavan Srinivasan, Venkat Rao Bagalkote,
	linux-kernel, Michael Ellerman, Nicholas Piggin, Shrikanth Hegde,
	Hari Bathini



Le 17/11/2026 à 13:33, Ritesh Harjani a écrit :
> Mike Rapoport <rppt@kernel.org> writes:
> 
>> On Mon, Nov 16, 2026 at 04:11:07PM +0100, Christophe Leroy (CS GROUP) wrote:
>>> Hi Mike,
>>>
>>> Le 15/11/2026 à 09:29, Mike Rapoport a écrit :
>>>> On Fri, Nov 13, 2026 at 06:01:37PM +0530, Ritesh Harjani (IBM) wrote:
>>>>> On book3s/32, a kprobe on an instruction that can't be emulated and has
>>>>> to be single stepped out of line, crashes the kernel. For e.g. with the
>>>>> following cmds...
>>>>>       ~> echo 9 > /proc/sys/kernel/printk
>>>>>       ~> mount -t debugfs debugfs /sys/kernel/debug
>>>>>       ~> echo 'p:wx filldir64+0xb8' > /sys/kernel/debug/tracing/kprobe_events
>>>>>       ~> echo 1 > /sys/kernel/debug/tracing/events/kprobes/wx/enable
>>>>>       ~> echo 1 > /sys/kernel/debug/tracing/events/kprobes/enable
>>>>>
>>>>> ...we can see the following crash:
>>>>>       kernel tried to execute exec-protected page (f101a008) - exploit attempt? (uid: 0)
>>>>>       BUG: Unable to handle kernel instruction fetch
>>>>>       Faulting instruction address: 0xc02b919c
>>>>>       Vector: 400 (Instruction Access) at [f19a9d60]
>>>>>           pc: c02b919c: filldir64+0xb8/0x2dc
>>>>>           lr: c02b9148: filldir64+0x64/0x2dc
>>>>>           sp: f19a9e20
>>>>>          msr: 10029032
>>>>>         current = 0xc112e900
>>>>>           pid   = 102, comm = ls
>>>>>       enter ? for help
>>>>>       [f19a9e20] c1a8d420 (unreliable)
>>>>>       [f19a9e50] c02e4074 dcache_readdir+0x208/0x2a8
>>>>>       [f19a9e90] c02b8c5c iterate_dir+0x88/0x1fc
>>>>>       [f19a9ec0] c02b96e0 sys_getdents64+0x88/0x198
>>>>>       [f19a9f10] c0014400 system_call_exception+0xf8/0x20c
>>>>>       [f19a9f30] c00191ac ret_from_syscall+0x0/0x2c
>>>>>       ---- Exception: c00 (System Call) at a7596164
>>>>>       SP (af802310) is in userspace
>>>>>
>>>>> This happens because on book3s/32 the exec permission is per segment
>>>>> (256MB), not per page. The page tables have no hardware exec bit, only
>>>>> the segment registers have a no-execute (N) bit. mmu_mark_initmem_nx()
>>>>> sets N on all the kernel segments except the ones of the modules area.
>>>>> So nothing in the vmalloc area can be executed, even when the page is
>>>>> mapped with PAGE_KERNEL_ROX or PAGE_KERNEL_EXEC, because _PAGE_EXEC is
>>>>> a software only bit there and the segment's N bit is what the MMU
>>>>> enforces.
>>>>>
>>>>> Before commit 1b750c2fbf82 ("powerpc: extend execmem_params for kprobes
>>>>> allocations"), the kprobes insn pages came from module_alloc(). That
>>>>> commit moved them to the whole of the vmalloc space, on the grounds
>>>>> that kprobes use a breakpoint and don't need to be near the kernel
>>>>> text. That is fine on platforms where the exec permission is per page,
>>>>> as PAGE_KERNEL_ROX is then enough to make a vmalloc page executable.
>>>>> It is not on book3s/32, where the modules area is not only close to
>>>>> the kernel text but is also the only page-mapped area whose segment is
>>>>> left executable. So the insn pages must keep coming from there.
>>>>>
>>>>> Hence, use the modules area for EXECMEM_KPROBES on book3s/32.
>>>> Right, I missed that on book3s/32 does vmalloc is not executable and its
>>>> protection can't be changed.
>>>>
>>>> Could it be that 8xx also need a special treatment for kprobes?
>>>
>>> Yes indeed, not for the same reasons but because 8xx ITLB-miss handler
>>> doesn't expect any instruction TLB miss on kernel addresses other than the
>>> ones from the module area:
>>>
>>> Commit 4afb0be7ccda ("powerpc/8xx: Avoid testing for kernel space in ITLB
>>> Miss.")
>>> Commit 16a71c045186 ("powerpc/8xx: Preallocate execmem page tables")
>>> Commit 33c527522f39 ("powerpc/8xx: Inconditionally use task PGDIR in ITLB
>>> misses")
>>>
> 
> Right. I looked at the above series which added all of this support [1].
> So looks like 603 might also have this related problem (other than the
> per segment N bit issue), since it's iTLB miss handler only walks the
> task's PGDIR, unlike it's dTLB handler which falls back to reloading the
> missing kernel PGD entry from swapper_pg_dir. So what I meant to say is
> that, similar to 8xx, a new PGD entry created in the kernel's vmalloc
> area for an insn page does not get copied over after the task is
> created (/PGDIR allocated), on 603 as well.
> But since MODLUES_VADDR is defined for both BOOK3S/32 and 8xx, so I guess it
> gets covered by the same change anyways.
> 
> [1]: https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Flore.kernel.org%2Flinuxppc-dev%2Fcover.1724173828.git.christophe.leroy%40csgroup.eu%2F&data=05%7C02%7Cchristophe.leroy%40csgroup.eu%7C0dc9911f0edd40821c2d08df44aea3d2%7C8b87af7d86474dc78df45f69a2011bb5%7C0%7C0%7C639305189016985969%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=UbPfu7OsJohLOnRi3b2v67TmweDvZBTi%2BCX3qcrp3qM%3D&reserved=0
> 
>>> So at the end as MODULES_VADDR/MODULES_END only exist on book3s/32 and 8xx I
>>> think when they exist they should be used as follows:
>>>
>>> diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
>>> index b617b69452cd..856f4609c488 100644
>>> --- a/arch/powerpc/mm/mem.c
>>> +++ b/arch/powerpc/mm/mem.c
>>> @@ -428,8 +428,13 @@ struct execmem_info __init *execmem_arch_setup(void)
>>>   				.fallback_end	= fallback_end,
>>>   			},
>>>   			[EXECMEM_KPROBES] = {
>>> +#ifdef MODULES_VADDR
>>> +				.start	= MODULES_VADDR,
>>> +				.end	= MODULES_END,
>>> +#else
>>>   				.start	= VMALLOC_START,
>>>   				.end	= VMALLOC_END,
>>> +#endif
>>
>> We already have #ifdef MODULES_VADDR a few lines above, I think the
>> assignment of kprobes_start and kprobes_end could go there instead of if
>> (IS_ENABLED(CONFIG_PPC_BOOK3S_32)).
>>
> 
> Right. So let me use the following diff then?
> 
> Please let me know if this diff looks good to you? If yes, then I can
> send a v3.

My preference would go to something similar to 'start' and 'end', no 
init at declaration, kprobes_start and kprobes_end set to 
VMALLOC_START/VMALLOC_END inside the #else

Christophe

> 
> Also - it will be good to know if patch-1 and patch-3 looks ok too,
> before I spin a next revision?
> 
> Thanks Christophe & Mike, for reviewing and pointing this out!
> 
> 
> diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
> index 77b63b52d8f2..547120a30bee 100644
> --- a/arch/powerpc/mm/mem.c
> +++ b/arch/powerpc/mm/mem.c
> @@ -391,6 +391,7 @@ struct execmem_info __init *execmem_arch_setup(void)
>          pgprot_t kprobes_prot = strict_kernel_rwx_enabled() ? PAGE_KERNEL_ROX : PAGE_KERNEL_EXEC;
>          pgprot_t prot = strict_module_rwx_enabled() ? PAGE_KERNEL : PAGE_KERNEL_EXEC;
>          unsigned long fallback_start = 0, fallback_end = 0;
> +       unsigned long kprobes_start = VMALLOC_START, kprobes_end = VMALLOC_END;
>          unsigned long start, end;
> 
>          /*
> @@ -410,6 +411,9 @@ struct execmem_info __init *execmem_arch_setup(void)
>          }
> 
>          end = MODULES_END;
> +
> +       kprobes_start = MODULES_VADDR;
> +       kprobes_end = MODULES_END;
>   #else
>          start = VMALLOC_START;
>          end = VMALLOC_END;
> @@ -428,8 +432,8 @@ struct execmem_info __init *execmem_arch_setup(void)
>                                  .fallback_end   = fallback_end,
>                          },
>                          [EXECMEM_KPROBES] = {
> -                               .start  = VMALLOC_START,
> -                               .end    = VMALLOC_END,
> +                               .start  = kprobes_start,
> +                               .end    = kprobes_end,
>                                  .pgprot = kprobes_prot,
>                                  .alignment = 1,
>                          },
> 
> 
> 
> -ritesh
> 
> 
>>>   				.pgprot	= kprobes_prot,
>>>   				.alignment = 1,
>>>   			},
>>>
>>>
>>>>
>>>>> Fixes: 1b750c2fbf82 ("powerpc: extend execmem_params for kprobes allocations")
>>>>> Signed-off-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
>>>>
>>>> Reviewed-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
>>>>
>>>>> ---
>>>>>    arch/powerpc/mm/mem.c | 10 ++++++++--
>>>>>    1 file changed, 8 insertions(+), 2 deletions(-)
>>>>>
>>>>> diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
>>>>> index 77b63b52d8f2..ae1bc7359353 100644
>>>>> --- a/arch/powerpc/mm/mem.c
>>>>> +++ b/arch/powerpc/mm/mem.c
>>>>> @@ -391,6 +391,7 @@ struct execmem_info __init *execmem_arch_setup(void)
>>>>>    	pgprot_t kprobes_prot = strict_kernel_rwx_enabled() ? PAGE_KERNEL_ROX : PAGE_KERNEL_EXEC;
>>>>>    	pgprot_t prot = strict_module_rwx_enabled() ? PAGE_KERNEL : PAGE_KERNEL_EXEC;
>>>>>    	unsigned long fallback_start = 0, fallback_end = 0;
>>>>> +	unsigned long kprobes_start = VMALLOC_START, kprobes_end = VMALLOC_END;
>>>>>    	unsigned long start, end;
>>>>>
>>>>>    	/*
>>>>> @@ -410,6 +411,11 @@ struct execmem_info __init *execmem_arch_setup(void)
>>>>>    	}
>>>>>
>>>>>    	end = MODULES_END;
>>>>> +
>>>>> +	if (IS_ENABLED(CONFIG_PPC_BOOK3S_32)) {
>>>>> +		kprobes_start = MODULES_VADDR;
>>>>> +		kprobes_end = MODULES_END;
>>>>> +	}
>>>>>    #else
>>>>>    	start = VMALLOC_START;
>>>>>    	end = VMALLOC_END;
>>>>> @@ -428,8 +434,8 @@ struct execmem_info __init *execmem_arch_setup(void)
>>>>>    				.fallback_end	= fallback_end,
>>>>>    			},
>>>>>    			[EXECMEM_KPROBES] = {
>>>>> -				.start	= VMALLOC_START,
>>>>> -				.end	= VMALLOC_END,
>>>>> +				.start	= kprobes_start,
>>>>> +				.end	= kprobes_end,
>>>>>    				.pgprot	= kprobes_prot,
>>>>>    				.alignment = 1,
>>>>>    			},
>>>>> --
>>>>> 2.39.5
>>>>>
>>>>>
>>>>
>>>
>>
>> -- 
>> Sincerely yours,
>> Mike.


^ permalink raw reply	[flat|nested] 14+ messages in thread

* Re: [PATCHv2 1/3] powerpc/execmem: Don't leave kprobes insn pages W+X when CONFIG_MODULES=n
  2026-11-13 12:31 ` [PATCHv2 1/3] powerpc/execmem: Don't leave kprobes insn pages W+X when CONFIG_MODULES=n Ritesh Harjani (IBM)
@ 2026-11-17 14:08   ` Christophe Leroy (CS GROUP)
  2026-11-17 17:15     ` Ritesh Harjani
  0 siblings, 1 reply; 14+ messages in thread
From: Christophe Leroy (CS GROUP) @ 2026-11-17 14:08 UTC (permalink / raw)
  To: Ritesh Harjani (IBM), linuxppc-dev
  Cc: Madhavan Srinivasan, Venkat Rao Bagalkote, linux-kernel,
	Michael Ellerman, Nicholas Piggin, Shrikanth Hegde, Hari Bathini



Le 13/11/2026 à 13:31, Ritesh Harjani (IBM) a écrit :
> commit 2e83e2ce34ff ("arch, mm: promote DEBUG_WX to CHECK_WX"), made
> CHECK_WX default STRICT_KERNEL_RWX.
> 
> Now on powerpc, when we boot with CONFIG_KPROBES=y and CONFIG_MODULES=n,
> we can see the following warning splat at boot:
> 
>      powerpc/mm: Found insecure W+X mapping at address 000000002ea08992/0xc008000000000000
>      WARNING: arch/powerpc/mm/ptdump/ptdump.c:194 at note_page+0x4dc/0x4fc, CPU#0: swapper/0/1
>      NIP [c0000000000ac204] note_page+0x4dc/0x4fc
>      LR [c0000000000ac200] note_page+0x4d8/0x4fc
>      Call Trace:
>       note_page+0x4d8/0x4fc (unreliable)
>       ptdump_pte_entry+0x70/0x94
>       walk_pte_range_inner+0x15c/0x260
>       walk_pgd_range+0x6d8/0xd5c
>       walk_page_range_debug+0xa4/0xf0
>       ptdump_walk_pgd+0xec/0x304
>       ptdump_check_wx+0xd8/0x154
>       kernel_init+0x94/0x26c
>       ret_from_kernel_user_thread+0x14/0x1c
>      Checked W+X mappings: FAILED, 1 W+X pages found
> 
> This happens because CHECK_WX noted a writeable + executable mapping.
> 
> This happens because execmem_arch_setup(), checks for
> struct_modules_rwx_enabled(), while kprobes depenency on modules was
> earlier removed via commit:
>      7582b7be16d0 ("kprobes: remove dependency on CONFIG_MODULES")
> Since the insn slots are only written with patch_instruction(), which
> does not need this mapping to be writable. This patch fixes it by
> checking for strict_kernel_rwx_enabled() instead.
> 
> Fixes: 7582b7be16d0 ("kprobes: remove dependency on CONFIG_MODULES")
> Signed-off-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
> ---
>   arch/powerpc/mm/mem.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
> index b617b69452cd..77b63b52d8f2 100644
> --- a/arch/powerpc/mm/mem.c
> +++ b/arch/powerpc/mm/mem.c
> @@ -388,7 +388,7 @@ static void prealloc_execmem_pgtable(void) { }
>   
>   struct execmem_info __init *execmem_arch_setup(void)
>   {
> -	pgprot_t kprobes_prot = strict_module_rwx_enabled() ? PAGE_KERNEL_ROX : PAGE_KERNEL_EXEC;
> +	pgprot_t kprobes_prot = strict_kernel_rwx_enabled() ? PAGE_KERNEL_ROX : PAGE_KERNEL_EXEC;
>   	pgprot_t prot = strict_module_rwx_enabled() ? PAGE_KERNEL : PAGE_KERNEL_EXEC;

Looking at this, I'd use PAGE_KERNEL_X instead of PAGE_KERNEL_EXEC.

Then we could get rid of PAGE_KERNEL_EXEC completely for powerpc.

>   	unsigned long fallback_start = 0, fallback_end = 0;
>   	unsigned long start, end;


^ permalink raw reply	[flat|nested] 14+ messages in thread

* Re: [PATCHv2 3/3] powerpc/ptdump: Show the exec flag of kernel mappings on e500
  2026-11-13 12:31 ` [PATCHv2 3/3] powerpc/ptdump: Show the exec flag of kernel mappings on e500 Ritesh Harjani (IBM)
@ 2026-11-17 14:09   ` Christophe Leroy (CS GROUP)
  0 siblings, 0 replies; 14+ messages in thread
From: Christophe Leroy (CS GROUP) @ 2026-11-17 14:09 UTC (permalink / raw)
  To: Ritesh Harjani (IBM), linuxppc-dev
  Cc: Madhavan Srinivasan, Venkat Rao Bagalkote, linux-kernel,
	Michael Ellerman, Nicholas Piggin, Shrikanth Hegde, Hari Bathini



Le 13/11/2026 à 13:31, Ritesh Harjani (IBM) a écrit :
> On e500, _PAGE_EXEC is made of two bits, _PAGE_BAP_SX and _PAGE_BAP_UX,
> and a kernel mapping only has _PAGE_BAP_SX (supervisor bit). The flag
> table requires all the bits of the mask to be set to print "X", so
> kernel_page_tables never shows a kernel mapping as executable:
> 
> w/o this patch:
>    0xf1015000-0xf1015fff  0x00000000         4K PTE   r          present  coherent                  accessed
> 
> Print "X" when any of the bits is set the same way it is done for
> _PAGE_READ and _PAGE_WRITE:
> 
> with this patch:
>    0xf1015000-0xf1015fff  0x00000000         4K PTE   r      X   present  coherent                  accessed
> 
> This only changes what is displayed. ptdump_check_wx() uses pte_exec()
> which already tests for either of the bits.
> 
> Signed-off-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>


Reviewed-by: Christophe Leroy (CS GROUP) <chleroy@kernel.org>



> ---
>   arch/powerpc/mm/ptdump/shared.c | 6 +++---
>   1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/powerpc/mm/ptdump/shared.c b/arch/powerpc/mm/ptdump/shared.c
> index edc69da19b85..e28a41a8b1ac 100644
> --- a/arch/powerpc/mm/ptdump/shared.c
> +++ b/arch/powerpc/mm/ptdump/shared.c
> @@ -22,9 +22,9 @@ static const struct flag_info flag_array[] = {
>   		.clear	= "w",
>   	}, {
>   		.mask	= _PAGE_EXEC,
> -		.val	= _PAGE_EXEC,
> -		.set	= " X ",
> -		.clear	= "   ",
> +		.val	= 0,
> +		.set	= "   ",
> +		.clear	= " X ",
>   	}, {
>   		.mask	= _PAGE_PRESENT,
>   		.val	= _PAGE_PRESENT,
> --
> 2.39.5
> 


^ permalink raw reply	[flat|nested] 14+ messages in thread

* Re: [PATCHv2 2/3] powerpc/execmem: Allocate kprobes insn pages from the modules area on book3s/32
  2026-11-17 14:04           ` Christophe Leroy (CS GROUP)
@ 2026-11-17 14:15             ` Mike Rapoport
  2026-11-17 17:33               ` Ritesh Harjani
  0 siblings, 1 reply; 14+ messages in thread
From: Mike Rapoport @ 2026-11-17 14:15 UTC (permalink / raw)
  To: Christophe Leroy (CS GROUP)
  Cc: Ritesh Harjani (IBM), linuxppc-dev, Madhavan Srinivasan,
	Venkat Rao Bagalkote, linux-kernel, Michael Ellerman,
	Nicholas Piggin, Shrikanth Hegde, Hari Bathini

On Tue, Nov 17, 2026 at 03:04:49PM +0100, Christophe Leroy (CS GROUP) wrote:
> Le 17/11/2026 à 13:33, Ritesh Harjani a écrit :
> > Mike Rapoport <rppt@kernel.org> writes:
> > 
> > > > > > Hence, use the modules area for EXECMEM_KPROBES on book3s/32.
> > > > > Right, I missed that on book3s/32 does vmalloc is not executable and its
> > > > > protection can't be changed.
> > > > > 
> > > > > Could it be that 8xx also need a special treatment for kprobes?
> > > > 
> > > > Yes indeed, not for the same reasons but because 8xx ITLB-miss handler
> > > > doesn't expect any instruction TLB miss on kernel addresses other than the
> > > > ones from the module area:
> > > > 
> > > > Commit 4afb0be7ccda ("powerpc/8xx: Avoid testing for kernel space in ITLB
> > > > Miss.")
> > > > Commit 16a71c045186 ("powerpc/8xx: Preallocate execmem page tables")
> > > > Commit 33c527522f39 ("powerpc/8xx: Inconditionally use task PGDIR in ITLB
> > > > misses")
> > 
> > Right. I looked at the above series which added all of this support [1].
> > So looks like 603 might also have this related problem (other than the
> > per segment N bit issue), since it's iTLB miss handler only walks the
> > task's PGDIR, unlike it's dTLB handler which falls back to reloading the
> > missing kernel PGD entry from swapper_pg_dir. So what I meant to say is
> > that, similar to 8xx, a new PGD entry created in the kernel's vmalloc
> > area for an insn page does not get copied over after the task is
> > created (/PGDIR allocated), on 603 as well.
> > But since MODLUES_VADDR is defined for both BOOK3S/32 and 8xx, so I guess it
> > gets covered by the same change anyways.
> > 
> > > > So at the end as MODULES_VADDR/MODULES_END only exist on book3s/32 and 8xx I
> > > > think when they exist they should be used as follows:
> > > > 
> > > > diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
> > > > index b617b69452cd..856f4609c488 100644
> > > > --- a/arch/powerpc/mm/mem.c
> > > > +++ b/arch/powerpc/mm/mem.c
> > > > @@ -428,8 +428,13 @@ struct execmem_info __init *execmem_arch_setup(void)
> > > >   				.fallback_end	= fallback_end,
> > > >   			},
> > > >   			[EXECMEM_KPROBES] = {
> > > > +#ifdef MODULES_VADDR
> > > > +				.start	= MODULES_VADDR,
> > > > +				.end	= MODULES_END,
> > > > +#else
> > > >   				.start	= VMALLOC_START,
> > > >   				.end	= VMALLOC_END,
> > > > +#endif
> > > 
> > > We already have #ifdef MODULES_VADDR a few lines above, I think the
> > > assignment of kprobes_start and kprobes_end could go there instead of if
> > > (IS_ENABLED(CONFIG_PPC_BOOK3S_32)).
> > > 
> > 
> > Right. So let me use the following diff then?
> > 
> > Please let me know if this diff looks good to you? If yes, then I can
> > send a v3.
> 
> My preference would go to something similar to 'start' and 'end', no init at
> declaration, kprobes_start and kprobes_end set to VMALLOC_START/VMALLOC_END
> inside the #else

I agree, it's more consistent this way.
 
> > Also - it will be good to know if patch-1 and patch-3 looks ok too,
> > before I spin a next revision?

Both make sense to me, feel free to add my

Acked-by: Mike Rapoport (Microsoft) <rppt@kernel.org>

-- 
Sincerely yours,
Mike.

^ permalink raw reply	[flat|nested] 14+ messages in thread

* Re: [PATCHv2 1/3] powerpc/execmem: Don't leave kprobes insn pages W+X when CONFIG_MODULES=n
  2026-11-17 14:08   ` Christophe Leroy (CS GROUP)
@ 2026-11-17 17:15     ` Ritesh Harjani
  0 siblings, 0 replies; 14+ messages in thread
From: Ritesh Harjani @ 2026-11-17 17:15 UTC (permalink / raw)
  To: Christophe Leroy (CS GROUP), linuxppc-dev
  Cc: Madhavan Srinivasan, Venkat Rao Bagalkote, linux-kernel,
	Michael Ellerman, Nicholas Piggin, Shrikanth Hegde, Hari Bathini

"Christophe Leroy (CS GROUP)" <chleroy@kernel.org> writes:

> Le 13/11/2026 à 13:31, Ritesh Harjani (IBM) a écrit :
>> commit 2e83e2ce34ff ("arch, mm: promote DEBUG_WX to CHECK_WX"), made
>> CHECK_WX default STRICT_KERNEL_RWX.
>> 
>> Now on powerpc, when we boot with CONFIG_KPROBES=y and CONFIG_MODULES=n,
>> we can see the following warning splat at boot:
>> 
>>      powerpc/mm: Found insecure W+X mapping at address 000000002ea08992/0xc008000000000000
>>      WARNING: arch/powerpc/mm/ptdump/ptdump.c:194 at note_page+0x4dc/0x4fc, CPU#0: swapper/0/1
>>      NIP [c0000000000ac204] note_page+0x4dc/0x4fc
>>      LR [c0000000000ac200] note_page+0x4d8/0x4fc
>>      Call Trace:
>>       note_page+0x4d8/0x4fc (unreliable)
>>       ptdump_pte_entry+0x70/0x94
>>       walk_pte_range_inner+0x15c/0x260
>>       walk_pgd_range+0x6d8/0xd5c
>>       walk_page_range_debug+0xa4/0xf0
>>       ptdump_walk_pgd+0xec/0x304
>>       ptdump_check_wx+0xd8/0x154
>>       kernel_init+0x94/0x26c
>>       ret_from_kernel_user_thread+0x14/0x1c
>>      Checked W+X mappings: FAILED, 1 W+X pages found
>> 
>> This happens because CHECK_WX noted a writeable + executable mapping.
>> 
>> This happens because execmem_arch_setup(), checks for
>> struct_modules_rwx_enabled(), while kprobes depenency on modules was
>> earlier removed via commit:
>>      7582b7be16d0 ("kprobes: remove dependency on CONFIG_MODULES")
>> Since the insn slots are only written with patch_instruction(), which
>> does not need this mapping to be writable. This patch fixes it by
>> checking for strict_kernel_rwx_enabled() instead.
>> 
>> Fixes: 7582b7be16d0 ("kprobes: remove dependency on CONFIG_MODULES")
>> Signed-off-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
>> ---
>>   arch/powerpc/mm/mem.c | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>> 
>> diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
>> index b617b69452cd..77b63b52d8f2 100644
>> --- a/arch/powerpc/mm/mem.c
>> +++ b/arch/powerpc/mm/mem.c
>> @@ -388,7 +388,7 @@ static void prealloc_execmem_pgtable(void) { }
>>   
>>   struct execmem_info __init *execmem_arch_setup(void)
>>   {
>> -	pgprot_t kprobes_prot = strict_module_rwx_enabled() ? PAGE_KERNEL_ROX : PAGE_KERNEL_EXEC;
>> +	pgprot_t kprobes_prot = strict_kernel_rwx_enabled() ? PAGE_KERNEL_ROX : PAGE_KERNEL_EXEC;
>>   	pgprot_t prot = strict_module_rwx_enabled() ? PAGE_KERNEL : PAGE_KERNEL_EXEC;
>
> Looking at this, I'd use PAGE_KERNEL_X instead of PAGE_KERNEL_EXEC.
>
> Then we could get rid of PAGE_KERNEL_EXEC completely for powerpc.
>

I think it's best if we keep the definition of PAGE_KERNEL_EXEC on
powerpc as PAGE_KERNEL_X, I think the core expects the arch to define
this, if not, then it defaults to PAGE_KERNEL e.g. see this:

include/linux/pgtable.h:
  #ifndef PAGE_KERNEL_EXEC
  # define PAGE_KERNEL_EXEC PAGE_KERNEL
  #endif

Also mm/execmem references PAGE_KERNEL_EXEC in __execmem_init().
Although it references only when execmem_arch_setup() is not defined by
the arch (which is not true for powerpc).


However, even though nothing breaks today, but later if there are other
users of PAGE_KERNEL_EXEC, then PAGE_KERNEL does not have X (execute) bit
enabled, so it might break in future. Plus looking at above, I believe
there is no harm in keeping it - since core expects the platform to
define it's own definition - due to all the other references it has.

Thoughts?

However, if you still think we should remove it (?) - then I feel let's
do that as a separate cleanup patch. Since this keeps the usage of EXEC
same between prot and kprobes_prot variables in patch-1, which we also
need to backport.


Thanks for the quick reviews!

-ritesh

^ permalink raw reply	[flat|nested] 14+ messages in thread

* Re: [PATCHv2 2/3] powerpc/execmem: Allocate kprobes insn pages from the modules area on book3s/32
  2026-11-17 14:15             ` Mike Rapoport
@ 2026-11-17 17:33               ` Ritesh Harjani
  0 siblings, 0 replies; 14+ messages in thread
From: Ritesh Harjani @ 2026-11-17 17:33 UTC (permalink / raw)
  To: Mike Rapoport, Christophe Leroy (CS GROUP)
  Cc: linuxppc-dev, Madhavan Srinivasan, Venkat Rao Bagalkote,
	linux-kernel, Michael Ellerman, Nicholas Piggin, Shrikanth Hegde,
	Hari Bathini

Mike Rapoport <rppt@kernel.org> writes:

> On Tue, Nov 17, 2026 at 03:04:49PM +0100, Christophe Leroy (CS GROUP) wrote:
>> Le 17/11/2026 à 13:33, Ritesh Harjani a écrit :
>> > Mike Rapoport <rppt@kernel.org> writes:
>> > 
>> > > > > > Hence, use the modules area for EXECMEM_KPROBES on book3s/32.
>> > > > > Right, I missed that on book3s/32 does vmalloc is not executable and its
>> > > > > protection can't be changed.
>> > > > > 
>> > > > > Could it be that 8xx also need a special treatment for kprobes?
>> > > > 
>> > > > Yes indeed, not for the same reasons but because 8xx ITLB-miss handler
>> > > > doesn't expect any instruction TLB miss on kernel addresses other than the
>> > > > ones from the module area:
>> > > > 
>> > > > Commit 4afb0be7ccda ("powerpc/8xx: Avoid testing for kernel space in ITLB
>> > > > Miss.")
>> > > > Commit 16a71c045186 ("powerpc/8xx: Preallocate execmem page tables")
>> > > > Commit 33c527522f39 ("powerpc/8xx: Inconditionally use task PGDIR in ITLB
>> > > > misses")
>> > 
>> > Right. I looked at the above series which added all of this support [1].
>> > So looks like 603 might also have this related problem (other than the
>> > per segment N bit issue), since it's iTLB miss handler only walks the
>> > task's PGDIR, unlike it's dTLB handler which falls back to reloading the
>> > missing kernel PGD entry from swapper_pg_dir. So what I meant to say is
>> > that, similar to 8xx, a new PGD entry created in the kernel's vmalloc
>> > area for an insn page does not get copied over after the task is
>> > created (/PGDIR allocated), on 603 as well.
>> > But since MODLUES_VADDR is defined for both BOOK3S/32 and 8xx, so I guess it
>> > gets covered by the same change anyways.
>> > 
>> > > > So at the end as MODULES_VADDR/MODULES_END only exist on book3s/32 and 8xx I
>> > > > think when they exist they should be used as follows:
>> > > > 
>> > > > diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
>> > > > index b617b69452cd..856f4609c488 100644
>> > > > --- a/arch/powerpc/mm/mem.c
>> > > > +++ b/arch/powerpc/mm/mem.c
>> > > > @@ -428,8 +428,13 @@ struct execmem_info __init *execmem_arch_setup(void)
>> > > >   				.fallback_end	= fallback_end,
>> > > >   			},
>> > > >   			[EXECMEM_KPROBES] = {
>> > > > +#ifdef MODULES_VADDR
>> > > > +				.start	= MODULES_VADDR,
>> > > > +				.end	= MODULES_END,
>> > > > +#else
>> > > >   				.start	= VMALLOC_START,
>> > > >   				.end	= VMALLOC_END,
>> > > > +#endif
>> > > 
>> > > We already have #ifdef MODULES_VADDR a few lines above, I think the
>> > > assignment of kprobes_start and kprobes_end could go there instead of if
>> > > (IS_ENABLED(CONFIG_PPC_BOOK3S_32)).
>> > > 
>> > 
>> > Right. So let me use the following diff then?
>> > 
>> > Please let me know if this diff looks good to you? If yes, then I can
>> > send a v3.
>> 
>> My preference would go to something similar to 'start' and 'end', no init at
>> declaration, kprobes_start and kprobes_end set to VMALLOC_START/VMALLOC_END
>> inside the #else

Sure.

>
> I agree, it's more consistent this way.
>  

Ok, then will go with the other approach.

>> > Also - it will be good to know if patch-1 and patch-3 looks ok too,
>> > before I spin a next revision?
>
> Both make sense to me, feel free to add my
>
> Acked-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
>

Thanks Mike and Christophe. Will make the changes in v3 as discussed.

-ritesh

^ permalink raw reply	[flat|nested] 14+ messages in thread

end of thread, other threads:[~2026-11-17 17:36 UTC | newest]

Thread overview: 14+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-11-13 12:31 [PATCHv2 0/3] powerpc: kprobes insn page fixes Ritesh Harjani (IBM)
2026-11-13 12:31 ` [PATCHv2 1/3] powerpc/execmem: Don't leave kprobes insn pages W+X when CONFIG_MODULES=n Ritesh Harjani (IBM)
2026-11-17 14:08   ` Christophe Leroy (CS GROUP)
2026-11-17 17:15     ` Ritesh Harjani
2026-11-13 12:31 ` [PATCHv2 2/3] powerpc/execmem: Allocate kprobes insn pages from the modules area on book3s/32 Ritesh Harjani (IBM)
2026-11-15  8:29   ` Mike Rapoport
2026-11-16 15:11     ` Christophe Leroy (CS GROUP)
2026-11-16 15:57       ` Mike Rapoport
2026-11-17 12:33         ` Ritesh Harjani
2026-11-17 14:04           ` Christophe Leroy (CS GROUP)
2026-11-17 14:15             ` Mike Rapoport
2026-11-17 17:33               ` Ritesh Harjani
2026-11-13 12:31 ` [PATCHv2 3/3] powerpc/ptdump: Show the exec flag of kernel mappings on e500 Ritesh Harjani (IBM)
2026-11-17 14:09   ` Christophe Leroy (CS GROUP)

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox