Message-ID: <0f838e71-429c-4029-8aba-f6389f6bba7d@redhat.com>
Date: Sun, 4 May 2025 10:56:46 +0200
From: Hans de Goede <hdegoede@...hat.com>
To: Mateusz Schyboll <dragonn@...pl>, Wentong Wu <wentong.wu@...el.com>,
 Alexander Usyskin <alexander.usyskin@...el.com>,
 Arnd Bergmann <arnd@...db.de>,
 Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Cc: Sakari Ailus <sakari.ailus@...ux.intel.com>,
 Stanislaw Gruszka <stanislaw.gruszka@...ux.intel.com>,
 linux-kernel@...r.kernel.org, stable@...nel.org
Subject: Re: [PATCH 1/2] mei: vsc: Fix fortify-panic caused by invalid
 counted_by() use

Hi Mateusz,

On 4-May-25 10:25 AM, Mateusz Schyboll wrote:
> Hi,
> 
> found this patch while searching for a I think a similar bug on 6.14.4 compiled with gcc 15:
> 
> [  121.966267] memcpy: detected buffer overflow: 3 byte read of buffer size 18446744073709551615

<snip.

> [  121.966409] Call Trace:
> [  121.966411]  <TASK>
> [  121.966412]  __fortify_panic+0xd/0xf
> [  121.966416]  bch2_xattr_get_trans.cold+0xe/0xe [bcachefs b5391e24358551d6754989dda9ec90add2ece543]

<snip>

> Unfortunately even with this patch the bug still shows up, so I am not 100% sure this is exactly the same , the call trace is diffirent so maybe a similar bug but in a diffirent subsystem?

That is not surprising, this really is a totally different bug,
which happens to also trigger a fortify panic and that you're seeing
this bug now that arch has switched to gcc15 indicates that it indeed
is triggered by gcc15 now supporting counted_by().

This could either be a real bug in the kernel which now gets caught or
another invalid counted_by() use.

Either way looking at the back trace it seems that you are using
bcachefs and this is a bcache fs issue (bch2_xattr_get_trans()
is defined in fs/bcachefs/xattr.c).

Please report this to the bcachefs developer. Note it would be
good to note in your report to Kent that this is triggered by gcc15.

Regards,

Hans





> The behavior from user space is that I am trying to launch Tiny Tina's Wonderlands and that pops up after the optimizing shaders finishes and the game never launches, it gets stuck after that.
> 
> W dniu 18.03.2025 o 15:12, Hans de Goede pisze:
>> gcc 15 honors the __counted_by(len) attribute on vsc_tp_packet.buf[]
>> and the vsc-tp.c code is using this in a wrong way. len does not contain
>> the available size in the buffer, it contains the actual packet length
>> *without* the crc. So as soon as vsc_tp_xfer() tries to add the crc to
>> buf[] the fortify-panic handler gets triggered:
>>
>> [   80.842193] memcpy: detected buffer overflow: 4 byte write of buffer size 0
>> [   80.842243] WARNING: CPU: 4 PID: 272 at lib/string_helpers.c:1032 __fortify_report+0x45/0x50
>> ...
>> [   80.843175]  __fortify_panic+0x9/0xb
>> [   80.843186]  vsc_tp_xfer.cold+0x67/0x67 [mei_vsc_hw]
>> [   80.843210]  ? seqcount_lockdep_reader_access.constprop.0+0x82/0x90
>> [   80.843229]  ? lockdep_hardirqs_on+0x7c/0x110
>> [   80.843250]  mei_vsc_hw_start+0x98/0x120 [mei_vsc]
>> [   80.843270]  mei_reset+0x11d/0x420 [mei]
>>
>> The easiest fix would be to just drop the counted-by but with the exception
>> of the ack buffer in vsc_tp_xfer_helper() which only contains enough room
>> for the packet-header, all other uses of vsc_tp_packet always use a buffer
>> of VSC_TP_MAX_XFER_SIZE bytes for the packet.
>>
>> Instead of just dropping the counted-by, split the vsc_tp_packet struct
>> definition into a header and a full-packet definition and use a fixed
>> size buf[] in the packet definition, this way fortify-source buffer
>> overrun checking still works when enabled.
>>
>> Fixes: 566f5ca97680 ("mei: Add transport driver for IVSC device")
>> Cc: stable@...nel.org
>> Signed-off-by: Hans de Goede <hdegoede@...hat.com>
>> ---
>>   drivers/misc/mei/vsc-tp.c | 26 +++++++++++++++-----------
>>   1 file changed, 15 insertions(+), 11 deletions(-)
>>
>> diff --git a/drivers/misc/mei/vsc-tp.c b/drivers/misc/mei/vsc-tp.c
>> index 7be1649b1972..fa553d4914b6 100644
>> --- a/drivers/misc/mei/vsc-tp.c
>> +++ b/drivers/misc/mei/vsc-tp.c
>> @@ -36,20 +36,24 @@
>>   #define VSC_TP_XFER_TIMEOUT_BYTES        700
>>   #define VSC_TP_PACKET_PADDING_SIZE        1
>>   #define VSC_TP_PACKET_SIZE(pkt) \
>> -    (sizeof(struct vsc_tp_packet) + le16_to_cpu((pkt)->len) + VSC_TP_CRC_SIZE)
>> +    (sizeof(struct vsc_tp_packet_hdr) + le16_to_cpu((pkt)->hdr.len) + VSC_TP_CRC_SIZE)
>>   #define VSC_TP_MAX_PACKET_SIZE \
>> -    (sizeof(struct vsc_tp_packet) + VSC_TP_MAX_MSG_SIZE + VSC_TP_CRC_SIZE)
>> +    (sizeof(struct vsc_tp_packet_hdr) + VSC_TP_MAX_MSG_SIZE + VSC_TP_CRC_SIZE)
>>   #define VSC_TP_MAX_XFER_SIZE \
>>       (VSC_TP_MAX_PACKET_SIZE + VSC_TP_XFER_TIMEOUT_BYTES)
>>   #define VSC_TP_NEXT_XFER_LEN(len, offset) \
>> -    (len + sizeof(struct vsc_tp_packet) + VSC_TP_CRC_SIZE - offset + VSC_TP_PACKET_PADDING_SIZE)
>> +    (len + sizeof(struct vsc_tp_packet_hdr) + VSC_TP_CRC_SIZE - offset + VSC_TP_PACKET_PADDING_SIZE)
>>   -struct vsc_tp_packet {
>> +struct vsc_tp_packet_hdr {
>>       __u8 sync;
>>       __u8 cmd;
>>       __le16 len;
>>       __le32 seq;
>> -    __u8 buf[] __counted_by(len);
>> +};
>> +
>> +struct vsc_tp_packet {
>> +    struct vsc_tp_packet_hdr hdr;
>> +    __u8 buf[VSC_TP_MAX_XFER_SIZE - sizeof(struct vsc_tp_packet_hdr)];
>>   };
>>     struct vsc_tp {
>> @@ -158,12 +162,12 @@ static int vsc_tp_dev_xfer(struct vsc_tp *tp, void *obuf, void *ibuf, size_t len
>>   static int vsc_tp_xfer_helper(struct vsc_tp *tp, struct vsc_tp_packet *pkt,
>>                     void *ibuf, u16 ilen)
>>   {
>> -    int ret, offset = 0, cpy_len, src_len, dst_len = sizeof(struct vsc_tp_packet);
>> +    int ret, offset = 0, cpy_len, src_len, dst_len = sizeof(struct vsc_tp_packet_hdr);
>>       int next_xfer_len = VSC_TP_PACKET_SIZE(pkt) + VSC_TP_XFER_TIMEOUT_BYTES;
>>       u8 *src, *crc_src, *rx_buf = tp->rx_buf;
>>       int count_down = VSC_TP_MAX_XFER_COUNT;
>>       u32 recv_crc = 0, crc = ~0;
>> -    struct vsc_tp_packet ack;
>> +    struct vsc_tp_packet_hdr ack;
>>       u8 *dst = (u8 *)&ack;
>>       bool synced = false;
>>   @@ -280,10 +284,10 @@ int vsc_tp_xfer(struct vsc_tp *tp, u8 cmd, const void *obuf, size_t olen,
>>         guard(mutex)(&tp->mutex);
>>   -    pkt->sync = VSC_TP_PACKET_SYNC;
>> -    pkt->cmd = cmd;
>> -    pkt->len = cpu_to_le16(olen);
>> -    pkt->seq = cpu_to_le32(++tp->seq);
>> +    pkt->hdr.sync = VSC_TP_PACKET_SYNC;
>> +    pkt->hdr.cmd = cmd;
>> +    pkt->hdr.len = cpu_to_le16(olen);
>> +    pkt->hdr.seq = cpu_to_le32(++tp->seq);
>>       memcpy(pkt->buf, obuf, olen);
>>         crc = ~crc32(~0, (u8 *)pkt, sizeof(pkt) + olen);
> 

Powered by blists - more mailing lists