linux-kernel - Re: [PATCH V2 XRT Alveo Infrastructure 3/9] of: handle fdt buffer alignment inside unflatten function

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: <e76c91d8-f2a2-f11b-14cb-da554bf42793@xilinx.com>
Date:   Tue, 30 Nov 2021 13:13:46 -0800
From:   Lizhi Hou <lizhi.hou@...inx.com>
To:     Rob Herring <robh@...nel.org>, Lizhi Hou <lizhi.hou@...inx.com>
CC:     <linux-kernel@...r.kernel.org>, <linux-fpga@...r.kernel.org>,
        <maxz@...inx.com>, <sonal.santan@...inx.com>, <yliu@...inx.com>,
        <michal.simek@...inx.com>, <stefanos@...inx.com>,
        <devicetree@...r.kernel.org>, <trix@...hat.com>, <mdf@...nel.org>,
        <dwmw2@...radead.org>, Max Zhen <max.zhen@...inx.com>
Subject: Re: [PATCH V2 XRT Alveo Infrastructure 3/9] of: handle fdt buffer
 alignment inside unflatten function


On 11/29/21 5:57 PM, Rob Herring wrote:
> On Fri, Nov 19, 2021 at 02:24:06PM -0800, Lizhi Hou wrote:
>> Add alignment check to of_fdt_unflatten_tree(). If it is not aligned,
>> allocate a aligned buffer and copy the fdt blob. So the caller does not
>> have to deal with the buffer alignment before calling this function.
>> XRT uses this function to unflatten fdt which is from Alveo firmware.
>>
>> Signed-off-by: Sonal Santan <sonal.santan@...inx.com>
>> Signed-off-by: Max Zhen <max.zhen@...inx.com>
>> Signed-off-by: Lizhi Hou <lizhi.hou@...inx.com>
>> ---
>>   drivers/of/fdt.c | 17 ++++++++++++++++-
>>   1 file changed, 16 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/of/fdt.c b/drivers/of/fdt.c
>> index 4546572af24b..d64445e43ceb 100644
>> --- a/drivers/of/fdt.c
>> +++ b/drivers/of/fdt.c
>> @@ -455,13 +455,28 @@ void *of_fdt_unflatten_tree(const unsigned long *blob,
>>                            struct device_node *dad,
>>                            struct device_node **mynodes)
>>   {
>> +     void *new_fdt = NULL, *fdt_align;
>>        void *mem;
>>
>> +     if (fdt_check_header(blob)) {
>> +             pr_err("Invalid fdt blob\n");
>> +             return NULL;
>> +     }
>> +     fdt_align = (void *)PTR_ALIGN(blob, FDT_ALIGN_SIZE);
>> +     if (fdt_align != blob) {
>> +             new_fdt = kmalloc(fdt_totalsize(blob) + FDT_ALIGN_SIZE, GFP_KERNEL);
>> +             if (!new_fdt)
>> +                     return NULL;
>> +             fdt_align = PTR_ALIGN(new_fdt, FDT_ALIGN_SIZE);
> Where's the copy?
>
>> +     }
>> +
>>        mutex_lock(&of_fdt_unflatten_mutex);
>> -     mem = __unflatten_device_tree(blob, dad, mynodes, &kernel_tree_alloc,
>> +     mem = __unflatten_device_tree(fdt_align, dad, mynodes, &kernel_tree_alloc,
>>                                      true);
>>        mutex_unlock(&of_fdt_unflatten_mutex);
>>
>> +     kfree(new_fdt);
> You know the unflattened DT just references strings and property values
> from the flattened DT. So you just caused a use after free.
>
> Fix your firmware to align the DT.

Thanks a lot for pointing this out. I did not aware the direct reference 
of strings and values from the flattened DT. And I will fix this.


Lizhi

>
> Rob