2018-12-14 15:25:42

by Christophe Leroy

Subject: [PATCH v2] powerpc/mm: make NULL pointer deferences explicit on bad page faults.

As several other arches including x86, this patch makes it explicit
that a bad page fault is a NULL pointer dereference when the fault
address is lower than PAGE_SIZE

In the mean time, this page makes all bad_page_fault() messages shorter
so that they remain on one single line. And it prefixes them by "BUG: "
so that they get easily greped.

Signed-off-by: Christophe Leroy <[email protected]>
---
v2: Taking into account comments from Michael

arch/powerpc/mm/fault.c | 19 +++++++++++--------
1 file changed, 11 insertions(+), 8 deletions(-)

diff --git a/arch/powerpc/mm/fault.c b/arch/powerpc/mm/fault.c
index 01b9bcc7fa85..3398291f4785 100644
--- a/arch/powerpc/mm/fault.c
+++ b/arch/powerpc/mm/fault.c
@@ -636,21 +636,24 @@ void bad_page_fault(struct pt_regs *regs, unsigned long address, int sig)
switch (TRAP(regs)) {
case 0x300:
case 0x380:
- printk(KERN_ALERT "Unable to handle kernel paging request for "
- "data at address 0x%08lx\n", regs->dar);
+ if (regs->dar < PAGE_SIZE)
+ pr_alert("BUG: Kernel NULL pointer dereference");
+ else
+ pr_alert("BUG: Unable to handle kernel data access");
+ pr_cont(" at 0x%08lx\n", regs->dar);
break;
case 0x400:
case 0x480:
- printk(KERN_ALERT "Unable to handle kernel paging request for "
- "instruction fetch\n");
+ pr_alert("BUG: Unable to handle kernel instruction fetch%s",
+ regs->nip < PAGE_SIZE ? " (NULL pointer ?)\n" : "\n");
break;
case 0x600:
- printk(KERN_ALERT "Unable to handle kernel paging request for "
- "unaligned access at address 0x%08lx\n", regs->dar);
+ pr_alert("BUG: Unable to handle kernel unaligned access at 0x%08lx\n",
+ regs->dar);
break;
default:
- printk(KERN_ALERT "Unable to handle kernel paging request for "
- "unknown fault\n");
+ pr_alert("BUG: Unable to handle unknown paging fault at 0x%08lx\n",
+ regs->dar);
break;
}
printk(KERN_ALERT "Faulting instruction address: 0x%08lx\n",
--
2.13.3

2018-12-17 17:01:39

by Michael Ellerman

Subject: Re: [PATCH v2] powerpc/mm: make NULL pointer deferences explicit on bad page faults.

Christophe Leroy <[email protected]> writes:

> diff --git a/arch/powerpc/mm/fault.c b/arch/powerpc/mm/fault.c
> index 01b9bcc7fa85..3398291f4785 100644
> --- a/arch/powerpc/mm/fault.c
> +++ b/arch/powerpc/mm/fault.c
> @@ -636,21 +636,24 @@ void bad_page_fault(struct pt_regs *regs, unsigned long address, int sig)
> switch (TRAP(regs)) {
> case 0x300:
> case 0x380:
> - printk(KERN_ALERT "Unable to handle kernel paging request for "
> - "data at address 0x%08lx\n", regs->dar);
> + if (regs->dar < PAGE_SIZE)
> + pr_alert("BUG: Kernel NULL pointer dereference");
> + else
> + pr_alert("BUG: Unable to handle kernel data access");
> + pr_cont(" at 0x%08lx\n", regs->dar);

It's best to avoid pr_cont() as it can lead to interleaving, so I
rewrote this as:

pr_alert("BUG: %s at 0x%08lx\n",
regs->dar < PAGE_SIZE ? "Kernel NULL pointer dereference" :
"Unable to handle kernel data access", regs->dar);


> break;
> case 0x400:
> case 0x480:
> - printk(KERN_ALERT "Unable to handle kernel paging request for "
> - "instruction fetch\n");
> + pr_alert("BUG: Unable to handle kernel instruction fetch%s",
> + regs->nip < PAGE_SIZE ? " (NULL pointer ?)\n" : "\n");
I dropped the space here ^


cheers

2018-12-23 11:13:33

by Michael Ellerman

Subject: Re: [v2] powerpc/mm: make NULL pointer deferences explicit on bad page faults.

On Fri, 2018-12-14 at 15:23:33 UTC, Christophe Leroy wrote:
> As several other arches including x86, this patch makes it explicit
> that a bad page fault is a NULL pointer dereference when the fault
> address is lower than PAGE_SIZE
>
> In the mean time, this page makes all bad_page_fault() messages shorter
> so that they remain on one single line. And it prefixes them by "BUG: "
> so that they get easily greped.
>
> Signed-off-by: Christophe Leroy <[email protected]>

Applied to powerpc next, thanks.

https://git.kernel.org/powerpc/c/49a502ea23bf9dec47f8f3c3960909

cheers