linux-kernel - Re: [PATCH] x86/boot: minor improvement in kaslr

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <7463228f-3749-8cab-c727-433bba4131b4@cn.fujitsu.com>
Date:   Fri, 1 Feb 2019 17:41:50 +0800
From:   Cao jin <caoj.fnst@...fujitsu.com>
To:     Kees Cook <keescook@...omium.org>
CC:     LKML <linux-kernel@...r.kernel.org>, X86 ML <x86@...nel.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
        "H. Peter Anvin" <hpa@...or.com>, Baoquan He <bhe@...hat.com>,
        Chao Fan <fanc.fnst@...fujitsu.com>
Subject: Re: [PATCH] x86/boot: minor improvement in kaslr

On 2/1/19 4:20 PM, Kees Cook wrote:
> On Fri, Feb 1, 2019 at 6:51 PM Cao jin <caoj.fnst@...fujitsu.com> wrote:
>>
>> comments fix: input_size is ZO image size which just don't count .bss
>> in, but has .text, .data, etc;
>> drop unecessary alignment: minimum is either 512M or output, both are
>> CONFIG_PHYSICAL_ALIGN aligned(output is aligned in head_32/64.S). But
>> mention it in earlier comments.
>>
>> Signed-off-by: Cao jin <caoj.fnst@...fujitsu.com>
>> ---
>>  arch/x86/boot/compressed/kaslr.c | 9 +++------
>>  1 file changed, 3 insertions(+), 6 deletions(-)
>>
>> diff --git a/arch/x86/boot/compressed/kaslr.c b/arch/x86/boot/compressed/kaslr.c
>> index 9ed9709d9947..a947c5aba34e 100644
>> --- a/arch/x86/boot/compressed/kaslr.c
>> +++ b/arch/x86/boot/compressed/kaslr.c
>> @@ -360,7 +360,7 @@ static void handle_mem_options(void)
>>   * (i.e. it does not include its run size). This range must be avoided
>>   * because it contains the data used for decompression.
>>   *
>> - * [input+input_size, output+init_size) is [_text, _end) for ZO. This
>> + * [input+input_size, output+init_size) is [_bss, _end) for ZO. This
> 
> This isn't right. The comment was correct before. See
> arch/x86/boot/compressed/vmlinux.lds.S for the layout of the ZO image:
> after the compressed image is _text, _rodata, _got, _data, _bss,
> _pgtable, and _end. "[_text, _end)" correctly identifies the span
> used.
> 

I am confused, doesn't input_size = ZO image size = .head.text +
.rodata..compressed + .rodata + .got + .data + .pgtable ? As I know,
.bss don't occupy any space in file, and ZO's full run size is against
the end of buffer, so I think the tiny gap here is just .bss, which is
also the stack and heap. Do I get it wrong?

>>   * range includes ZO's heap and stack, and must be avoided since it
>>   * performs the decompression.
>>   *
>> @@ -763,9 +763,6 @@ static unsigned long find_random_phys_addr(unsigned long minimum,
>>                 return 0;
>>         }
>>
>> -       /* Make sure minimum is aligned. */
>> -       minimum = ALIGN(minimum, CONFIG_PHYSICAL_ALIGN);
>> -
> 
> I would prefer to keep this runtime calculation since it enforces the
> requirement instead of making leaving it in a comment. When this goes
> wrong, you get an unbootable kernel, which is very frustrating to
> debug.
> 

I find that I maybe wrong here. It is said that CONFIG_PHYSICAL_ALIGN
must be a multiple of 0x200000 on 64-bit, so it could be 2M, 4M, 6M, 8M,
12M, 14M, 16M, while 512M can't be divided by 6, 10, 12, 14 without
remainder.

-- 
Sincerely,
Cao jin