linux-kernel - Re: [PATCH v2 2/3] PCI: Prevent overflow in proc_bus_pci

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] [day] [month] [year] [list]

Message-ID: <6c63e3b4-c542-4a9f-bc9f-fa214a139039@huawei.com>
Date: Tue, 30 Dec 2025 16:20:32 +0800
From: duziming <duziming2@...wei.com>
To: Bjorn Helgaas <helgaas@...nel.org>
CC: <bhelgaas@...gle.com>, <jbarnes@...tuousgeek.org>, <chrisw@...hat.com>,
	<alex.williamson@...hat.com>, <linux-pci@...r.kernel.org>,
	<linux-kernel@...r.kernel.org>, <liuyongqiang13@...wei.com>,
	Krzysztof Wilczyński <kwilczynski@...nel.org>
Subject: Re: [PATCH v2 2/3] PCI: Prevent overflow in proc_bus_pci_write()


在 2025/12/30 2:07, Bjorn Helgaas 写道:
> [+cc Krzysztof; I thought we looked at this long ago?]
>
> On Wed, Dec 24, 2025 at 05:27:18PM +0800, Ziming Du wrote:
>> From: Yongqiang Liu <liuyongqiang13@...wei.com>
>>
>> When the value of ppos over the INT_MAX, the pos is over set to a negtive
>> value which will be passed to get_user() or pci_user_write_config_dword().
>> Unexpected behavior such as a softlock will happen as follows:
> s/negtive/negative/
> s/softlock/soft lockup/ to match message below
Thanks for pointing out the ambiguous parts.
> s/ppos/pos/ (or fix this to refer to "*ppos", which I think is what
> you're referring to)
>
> I guess the point is that proc_bus_pci_write() takes a "loff_t *ppos",
> loff_t is a signed type, and negative read/write offsets are invalid.

Actually, the *loff_t *ppos *passed in is not a negative value. The root 
cause of the issue

lies in the cast *int* *pos = *ppos*. When the value of **ppos* over the 
INT_MAX, the pos is over set

to a negative value. This negative *pos* then propagates through 
subsequent logic, leading to the observed errors.

> If this is easily reproducible with "dd" or similar, could maybe
> include a sample command line?

We reproduced the issue using the following POC:

     #include <stdio.h>

     #include <string.h>
     #include <unistd.h>
     #include <fcntl.h>
     #include <sys/uio.h>

     int main() {
     int fd = open("/proc/bus/pci/00/02.0", O_RDWR);
     if (fd < 0) {
         perror("open failed");
         return 1;
     }
     char data[] = "926b7719201054f37a1d9d391e862c";
     off_t offset = 0x80800001;
     struct iovec iov = {
         .iov_base = data,
         .iov_len = 0xf
     };
     pwritev(fd, &iov, 1, offset);
     return 0;
}

>>   watchdog: BUG: soft lockup - CPU#0 stuck for 130s! [syz.3.109:3444]
>>   RIP: 0010:_raw_spin_unlock_irq+0x17/0x30
>>   Call Trace:
>>    <TASK>
>>    pci_user_write_config_dword+0x126/0x1f0
>>    proc_bus_pci_write+0x273/0x470
>>    proc_reg_write+0x1b6/0x280
>>    do_iter_write+0x48e/0x790
>>    vfs_writev+0x125/0x4a0
>>    __x64_sys_pwritev+0x1e2/0x2a0
>>    do_syscall_64+0x59/0x110
>>    entry_SYSCALL_64_after_hwframe+0x78/0xe2
>>
>> Fix this by add check for the pos.
>>
>> Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
>> Signed-off-by: Yongqiang Liu <liuyongqiang13@...wei.com>
>> Signed-off-by: Ziming Du <duziming2@...wei.com>
>> ---
>>   drivers/pci/proc.c | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/pci/proc.c b/drivers/pci/proc.c
>> index 9348a0fb8084..200d42feafd8 100644
>> --- a/drivers/pci/proc.c
>> +++ b/drivers/pci/proc.c
>> @@ -121,7 +121,7 @@ static ssize_t proc_bus_pci_write(struct file *file, const char __user *buf,
>>   	if (ret)
>>   		return ret;
>>   
>> -	if (pos >= size)
>> +	if (pos >= size || pos < 0)
>>   		return 0;
> I see a few similar cases that look like this; maybe we should do the
> same?
>
>    if (pos < 0)
>      return -EINVAL;
>
> Looks like proc_bus_pci_read() has the same issue?

proc_bus_pci_read() may also trigger similar issue as mentioned by Ilpo 
Järvinen in

https://lore.kernel.org/linux-pci/e5a91378-4a41-32fb-00c6-2810084581bd@linux.intel.com/

However, it does not result in an overflow to a negative number.

>
> What about pci_read_config(), pci_write_config(),
> pci_llseek_resource(), pci_read_legacy_io(), pci_write_legacy_io(),
> pci_read_resource_io(), pci_write_resource_io(), pci_read_rom()?
> These are all sysfs things; does the sysfs infrastructure take care of
> negative offsets before we get to these?

In do_pwritev(), the following check has been performed:

    if (pos < 0)
          return -EINVAL;

Theoretically, a negative offset should not occur.

>>   	if (nbytes >= size)
>>   		nbytes = size;
>> -- 
>> 2.43.0
>>