Re: [PATCH v6 2/4] x86/syscalls: Specific usage of verify_pre_usermode_state

[H. Peter Anvin]

<will.deacon@xxxxxxx>,Catalin Marinas <catalin.marinas@xxxxxxx>,Mark Rutland <mark.rutland@xxxxxxx>,James Morse <james.morse@xxxxxxx>,linux-s390 <linux-s390@xxxxxxxxxxxxxxx>,LKML <linux-kernel@xxxxxxxxxxxxxxx>,Linux API <linux-api@xxxxxxxxxxxxxxx>,the arch/x86 maintainers <x86@xxxxxxxxxx>,linux-arm-kernel@xxxxxxxxxxxxxxxxxxx,Kernel Hardening <kernel-hardening@xxxxxxxxxxxxxxxxxx>
From: hpa@xxxxxxxxx
Message-ID: <C92745AD-C4D9-441E-854C-985512E5FD8F@xxxxxxxxx>

On April 4, 2017 12:21:48 PM PDT, Thomas Garnier <thgarnie@xxxxxxxxxx> wrote:
>On Tue, Apr 4, 2017 at 11:27 AM, H. Peter Anvin <hpa@xxxxxxxxx> wrote:
>> On 04/04/17 10:47, Thomas Garnier wrote:
>>> diff --git a/arch/x86/include/asm/pgtable_64_types.h
>b/arch/x86/include/asm/pgtable_64_types.h
>>> index 516593e66bd6..12fa851c7fa8 100644
>>> --- a/arch/x86/include/asm/pgtable_64_types.h
>>> +++ b/arch/x86/include/asm/pgtable_64_types.h
>>> @@ -78,4 +78,15 @@ typedef struct { pteval_t pte; } pte_t;
>>>
>>>  #define EARLY_DYNAMIC_PAGE_TABLES    64
>>>
>>> +/*
>>> + * User space process size. 47bits minus one guard page.  The guard
>>> + * page is necessary on Intel CPUs: if a SYSCALL instruction is at
>>> + * the highest possible canonical userspace address, then that
>>> + * syscall will enter the kernel with a non-canonical return
>>> + * address, and SYSRET will explode dangerously.  We avoid this
>>> + * particular problem by preventing anything from being mapped
>>> + * at the maximum canonical address.
>>> + */
>>> +#define TASK_SIZE_MAX        ((_AC(1, UL) << 47) - PAGE_SIZE)
>>> +
>>>  #endif /* _ASM_X86_PGTABLE_64_DEFS_H */
>>> diff --git a/arch/x86/include/asm/processor.h
>b/arch/x86/include/asm/processor.h
>>> index 3cada998a402..e80822582d3e 100644
>>> --- a/arch/x86/include/asm/processor.h
>>> +++ b/arch/x86/include/asm/processor.h
>>> @@ -825,17 +825,6 @@ static inline void spin_lock_prefetch(const
>void *x)
>>>  #define KSTK_ESP(task)               (task_pt_regs(task)->sp)
>>>
>>>  #else
>>> -/*
>>> - * User space process size. 47bits minus one guard page.  The guard
>>> - * page is necessary on Intel CPUs: if a SYSCALL instruction is at
>>> - * the highest possible canonical userspace address, then that
>>> - * syscall will enter the kernel with a non-canonical return
>>> - * address, and SYSRET will explode dangerously.  We avoid this
>>> - * particular problem by preventing anything from being mapped
>>> - * at the maximum canonical address.
>>> - */
>>> -#define TASK_SIZE_MAX        ((1UL << 47) - PAGE_SIZE)
>>> -
>>>  /* This decides where the kernel will search for a free chunk of vm
>>>   * space during mmap's.
>>>   */
>>>
>>
>> This should be an entirely separate patch; if nothing else you need
>to
>> explain it in the comments.
>
>I will explain it in the commit message, it should be easier than a
>separate patch.
>
>>
>> Also, you say this is for "x86", but I still don't see any code for
>i386
>> whatsoever.  Have you verified *all* the i386 and i386-compat paths
>to
>> make sure they go via prepare_exit_to_usermode()?  [Cc: Andy]
>
>I did but I will do it again for the next iteration.
>
>>
>> Finally, I can't really believe I'm the only person for whom
>"Specific
>> usage of verity_pre_usermode_state" is completely opaque.
>
>I agree, I will improve it.
>
>>
>>         -hpa
>>

Easier for you, perhaps, but not for everyone else...
-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.