Resetting device doesn't properly work

[Civil]

I've encountered that when I was trying to boot: tt-bh-linux

# make boot
/root/.tenstorrent-venv/bin/python boot.py --boot --l2cpu 0 --opensbi_bin fw_jump.bin --opensbi_dst 0x400030000000 --rootfs_bin rootfs.ext4 --rootfs_dst 0x4000e5000000 --kernel_bin Image --kernel_dst 0x400030200000 --dtb_bin blackhole-p100.dtb --dtb_dst 0x400030100000
 Starting PCI link reset on BH devices at PCI indices: 0
Waiting for up to 2 seconds for asic to come back after reset
Traceback (most recent call last):
  File "/root/tt/tt-bh-linux/boot.py", line 157, in <module>
    main()
    ~~~~^^
  File "/root/tt/tt-bh-linux/boot.py", line 84, in main
    pci_board_reset([0])
    ~~~~~~~~~~~~~~~^^^^^
  File "/root/.tenstorrent-venv/lib/python3.13/site-packages/tt_smi/tt_smi_backend.py", line 772, in pci_board_reset
    BHChipReset().full_lds_reset(pci_interfaces=reset_bh_pci_idx)
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/.tenstorrent-venv/lib/python3.13/site-packages/tt_tools_common/reset_common/bh_reset.py", line 136, in full_lds_reset
    command_memory_byte = os.pread(file.fileno(), 1, 4)
OSError: [Errno 19] No such device
make: *** [Makefile:80: boot] Error 1

[Civil]

It seems to me that root cause is that /sys/bus/pci/devices/<pci_id>/config might disappear after reset and therefore previous file handle won't be valid anymore.

[Civil]

I've actually needed to add 3 changes:

1. Increase timeout to 240s (that is a bit too much, but sometimes it didn't perform reset fast enough)
2. Add try-except and catch FileNotFound exception
3. In detect_chips add several tries with a sleep in between

[Civil]

#28 - that fixes all the problems for me, but I don't like that the whole reset logic rely on time and on system to be able to see the change in time.

[warthog9]

Double check with the latest reset changes that use the kmd, this might be sorted now