diff --git a/python/spi.py b/python/spi.py
index 89018a6a..d853ba80 100644
--- a/python/spi.py
+++ b/python/spi.py
@@ -322,6 +322,9 @@ class STBootloaderSPIHandle(BaseSTBootloaderHandle):
   """
     Implementation of the STM32 SPI bootloader protocol described in:
     https://www.st.com/resource/en/application_note/an4286-spi-protocol-used-in-the-stm32-bootloader-stmicroelectronics.pdf
+
+    NOTE: the bootloader's state machine is fragile and immediately gets into a bad state when
+          sending any junk, e.g. when using the panda SPI protocol.
   """
 
   SYNC = 0x5A
diff --git a/tests/hitl/5_spi.py b/tests/hitl/5_spi.py
index 6964c9ab..106a1c7d 100644
--- a/tests/hitl/5_spi.py
+++ b/tests/hitl/5_spi.py
@@ -3,29 +3,10 @@ import pytest
 import random
 from unittest.mock import patch
 
-from panda import Panda, PandaDFU
-from panda.python.spi import SpiDevice, PandaProtocolMismatch, PandaSpiNackResponse
+from panda import Panda
+from panda.python.spi import PandaProtocolMismatch, PandaSpiNackResponse
 
 
-@pytest.mark.skip("doesn't work, bootloader seems to ignore commands once it sees junk")
-def test_dfu_with_spam(p):
-  dfu_serial = p.get_dfu_serial()
-
-  # enter DFU
-  p.reset(enter_bootstub=True)
-  p.reset(enter_bootloader=True)
-  assert Panda.wait_for_dfu(dfu_serial, timeout=19), "failed to enter DFU"
-
-  # send junk
-  d = SpiDevice()
-  for _ in range(9):
-    with d.acquire() as spi:
-      dat = [random.randint(-1, 255) for _ in range(random.randint(1, 100))]
-      spi.xfer(dat)
-
-    # should still show up
-    assert dfu_serial in PandaDFU.list()
-
 class TestSpi:
   def _ping(self, mocker, panda):
     # should work with no retries