[systemd-devel] Weird udev issue: char device replaced by regular file on suspend

[Johannes Bauer]

Hi list,

I'm seeing a very odd issue with udev and I'm not really sure which
component could/would be responsible -- udev is pretty much my only hope.

Here's the situation: I have a USB/serial converter attached to my
computer. It's detected as /dev/ttyUSB0. A program is started as my
local user (UID 1000) and forks into the background. Every now and then
it sends one of four characters to this device and reads a bit of data
back. So far, so good, everything working as expected.

Now the odd thing: When I put my computer into suspend-to-RAM, I'm
seeing something very odd when the thing wakes back up (about 3 in 4
times this happens, nondeterministically). The character device
/dev/ttyUSB0 is replaced by a regular file with 0644 permissions which
is owned by root:root:

joequad joe [~]: ls /dev/ttyUSB0
-rw-r--r-- 1 root root 1   25.06.2015 18:09:29 /dev/ttyUSB0
joequad joe [~]: cat /dev/ttyUSB0 | hexdump -C
00000000  24                                                |$|
00000001

It's one byte long and this one byte is exactly the data that my
userspace program tried to send. But obviously my user space program
could not have created that node (no permission to write in /dev,
obviously, and my program has no suid or anything, runs just as UID 1000).

Here's what dmesg says when suspending/waking up:

[...] lots of output [...]
[186649.272605] ark3116 ttyUSB0: ark3116 converter now disconnected from
ttyUSB0
[186649.272619] ark3116 7-2.2:1.0: device disconnected
[186650.452061] ata7: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
[186650.888603] ACPI Error: [DSSP] Namespace lookup failure,
AE_NOT_FOUND (20131115/psargs-359)
[186650.888605] ACPI Error: Method parse/execution failed
[\_SB_.PCI0.SAT1.SPT2._GTF] (Node ffff880893e1ae38), AE_NOT_FOUND
(20131115/psparse-536)
[186651.030846] ACPI Error: [DSSP] Namespace lookup failure,
AE_NOT_FOUND (20131115/psargs-359)
[186651.030849] ACPI Error: Method parse/execution failed
[\_SB_.PCI0.SAT1.SPT2._GTF] (Node ffff880893e1ae38), AE_NOT_FOUND
(20131115/psparse-536)
[186651.031003] ata7.00: configured for UDMA/133
[186651.031097] sd 6:0:0:0: [sdc] Starting disk
[186652.988078] ata6: link is slow to respond, please be patient (ready=0)
[186655.676059] ata6: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
[186655.679062] ACPI Error: [DSSP] Namespace lookup failure,
AE_NOT_FOUND (20131115/psargs-359)
[186655.679069] ACPI Error: Method parse/execution failed
[\_SB_.PCI0.SAT1.SPT1._GTF] (Node ffff880893e1adc0), AE_NOT_FOUND
(20131115/psparse-536)
[186655.712350] ACPI Error: [DSSP] Namespace lookup failure,
AE_NOT_FOUND (20131115/psargs-359)
[186655.712356] ACPI Error: Method parse/execution failed
[\_SB_.PCI0.SAT1.SPT1._GTF] (Node ffff880893e1adc0), AE_NOT_FOUND
(20131115/psparse-536)
[186655.712521] ata6.00: configured for UDMA/133
[186655.712640] sd 5:0:0:0: [sdb] Starting disk
[186655.720875] PM: resume of devices complete after 8128.520 msecs
[186655.720895] ark3116 7-2.2:1.0: ark3116 converter detected
[186655.721740] usb 7-2.2: ark3116 using RS232 mode
[186655.721822] usb 7-2.2: ark3116 converter now attached to ttyUSB0
[186655.722165] PM: Finishing wakeup.

My userspace program happens to hang in a call to write(2) when the PC
is put in suspend. The write returns with errno = 5 (I/O error) when the
PC comes back up.

Another clue is here: I've seen this issue with my old (much slower PC)
before. But it happened very sporadically, about once every 30
suspend/wakeup-cycles. I thought this was odd, but blamed some weird
random component to do this. Now I've switched to a much faster computer
and am seeing this issue in around 75% of all suspend/wakeup-cycles and
it's really annoying the hell out of me. So I'm guessing there's some
kind of race condition involved which hits much more often now that my
PC is fast enough.

Oh, I'm running Linux Mint on x86_64. udev is version 204.

If you have any hints whatsoever on how I could debug this really weird
issue, please let me know. I really just don't have a clue how to debug
udev/this weird phaenomenon.

Thank you so much,
Best regards,
Johannes