All of lore.kernel.org
 help / color / mirror / Atom feed
From: Alexander Bulekov <alxndr@bu.edu>
To: qemu-devel@nongnu.org
Cc: Thomas Huth <thuth@redhat.com>, Alexander Bulekov <alxndr@bu.edu>,
	f4bug@amsat.org, darren.kenny@oracle.com, bsd@redhat.com,
	stefanha@redhat.com, pbonzini@redhat.com,
	dimastep@yandex-team.ru
Subject: [PATCH v5 10/16] scripts/oss-fuzz: Add script to reorder a general-fuzzer trace
Date: Tue, 20 Oct 2020 21:07:46 -0400	[thread overview]
Message-ID: <20201021010752.973230-11-alxndr@bu.edu> (raw)
In-Reply-To: <20201021010752.973230-1-alxndr@bu.edu>

The general-fuzzer uses hooks to fulfill DMA requests just-in-time.
This means that if we try to use QTEST_LOG=1 to build a reproducer, the
DMA writes will be logged _after_ the in/out/read/write that triggered
the DMA read. To work work around this, the general-fuzzer annotates
these just-in time DMA fulfilments with a tag that we can use to
discern them. This script simply iterates over a raw qtest
trace (including log messages, errors, timestamps etc), filters it and
re-orders it so that DMA fulfillments are placed directly _before_ the
qtest command that will cause the DMA access.

Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
---
 .../oss-fuzz/reorder_fuzzer_qtest_trace.py    | 103 ++++++++++++++++++
 1 file changed, 103 insertions(+)
 create mode 100755 scripts/oss-fuzz/reorder_fuzzer_qtest_trace.py

diff --git a/scripts/oss-fuzz/reorder_fuzzer_qtest_trace.py b/scripts/oss-fuzz/reorder_fuzzer_qtest_trace.py
new file mode 100755
index 0000000000..47295c9e69
--- /dev/null
+++ b/scripts/oss-fuzz/reorder_fuzzer_qtest_trace.py
@@ -0,0 +1,103 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+"""
+Use this to convert qtest log info from a generic fuzzer input into a qtest
+trace that you can feed into a standard qemu-system process. Example usage:
+
+QEMU_FUZZ_ARGS="-machine q35,accel=qtest" QEMU_FUZZ_OBJECTS="*" \
+        ./i386-softmmu/qemu-fuzz-i386 --fuzz-target=general-pci-fuzz
+# .. Finds some crash
+QTEST_LOG=1 FUZZ_SERIALIZE_QTEST=1 \
+QEMU_FUZZ_ARGS="-machine q35,accel=qtest" QEMU_FUZZ_OBJECTS="*" \
+        ./i386-softmmu/qemu-fuzz-i386 --fuzz-target=general-pci-fuzz
+        /path/to/crash 2> qtest_log_output
+scripts/oss-fuzz/reorder_fuzzer_qtest_trace.py qtest_log_output > qtest_trace
+./i386-softmmu/qemu-fuzz-i386 -machine q35,accel=qtest \
+        -qtest stdin < qtest_trace
+
+### Details ###
+
+Some fuzzer make use of hooks that allow us to populate some memory range, just
+before a DMA read from that range. This means that the fuzzer can produce
+activity that looks like:
+    [start] read from mmio addr
+    [end]   read from mmio addr
+    [start] write to pio addr
+        [start] fill a DMA buffer just in time
+        [end]   fill a DMA buffer just in time
+        [start] fill a DMA buffer just in time
+        [end]   fill a DMA buffer just in time
+    [end]   write to pio addr
+    [start] read from mmio addr
+    [end]   read from mmio addr
+
+We annotate these "nested" DMA writes, so with QTEST_LOG=1 the QTest trace
+might look something like:
+[R +0.028431] readw 0x10000
+[R +0.028434] outl 0xc000 0xbeef  # Triggers a DMA read from 0xbeef and 0xbf00
+[DMA][R +0.034639] write 0xbeef 0x2 0xAAAA
+[DMA][R +0.034639] write 0xbf00 0x2 0xBBBB
+[R +0.028431] readw 0xfc000
+
+This script would reorder the above trace so it becomes:
+readw 0x10000
+write 0xbeef 0x2 0xAAAA
+write 0xbf00 0x2 0xBBBB
+outl 0xc000 0xbeef
+readw 0xfc000
+
+I.e. by the time, 0xc000 tries to read from DMA, those DMA buffers have already
+been set up, removing the need for the DMA hooks. We can simply provide this
+reordered trace via -qtest stdio to reproduce the input
+
+Note: this won't work for traces where the device tries to read from the same
+DMA region twice in between MMIO/PIO commands. E.g:
+    [R +0.028434] outl 0xc000 0xbeef
+    [DMA][R +0.034639] write 0xbeef 0x2 0xAAAA
+    [DMA][R +0.034639] write 0xbeef 0x2 0xBBBB
+
+The fuzzer will annotate suspected double-fetches with [DOUBLE-FETCH]. This
+script looks for these tags and warns the users that the resulting trace might
+not reproduce the bug.
+"""
+
+import sys
+
+__author__     = "Alexander Bulekov <alxndr@bu.edu>"
+__copyright__  = "Copyright (C) 2020, Red Hat, Inc."
+__license__    = "GPL version 2 or (at your option) any later version"
+
+__maintainer__ = "Alexander Bulekov"
+__email__      = "alxndr@bu.edu"
+
+
+def usage():
+    sys.exit("Usage: {} /path/to/qtest_log_output".format((sys.argv[0])))
+
+
+def main(filename):
+    with open(filename, "r") as f:
+        trace = f.readlines()
+
+    # Leave only lines that look like logged qtest commands
+    trace[:] = [x.strip() for x in trace if "[R +" in x
+                or "[S +" in x and "CLOSED" not in x]
+
+    for i in range(len(trace)):
+        if i+1 < len(trace):
+            if "[DMA]" in trace[i+1]:
+                if "[DOUBLE-FETCH]" in trace[i+1]:
+                    sys.stderr.write("Warning: Likely double fetch on line"
+                                     "{}.\n There will likely be problems "
+                                     "reproducing behavior with the "
+                                     "resulting qtest trace\n\n".format(i+1))
+                trace[i], trace[i+1] = trace[i+1], trace[i]
+    for line in trace:
+        print(line.split("]")[-1].strip())
+
+
+if __name__ == '__main__':
+    if len(sys.argv) == 1:
+        usage()
+    main(sys.argv[1])
-- 
2.28.0



  parent reply	other threads:[~2020-10-21  1:13 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-10-21  1:07 [PATCH v5 00/16] Add a Generic Virtual Device Fuzzer Alexander Bulekov
2020-10-21  1:07 ` [PATCH v5 01/16] memory: Add FlatView foreach function Alexander Bulekov
2020-10-21  1:07 ` [PATCH v5 02/16] fuzz: Add general virtual-device fuzzer Alexander Bulekov
2020-10-21 12:44   ` Paolo Bonzini
2020-10-21 14:49     ` Alexander Bulekov
2020-10-21  1:07 ` [PATCH v5 03/16] fuzz: Add PCI features to the general fuzzer Alexander Bulekov
2020-10-21  1:07 ` [PATCH v5 04/16] fuzz: Add DMA support to the generic-fuzzer Alexander Bulekov
2020-10-21  1:07 ` [PATCH v5 05/16] fuzz: Declare DMA Read callback function Alexander Bulekov
2020-10-21  1:07 ` [PATCH v5 06/16] fuzz: Add fuzzer callbacks to DMA-read functions Alexander Bulekov
2020-10-21  1:07 ` [PATCH v5 07/16] fuzz: Add support for custom crossover functions Alexander Bulekov
2020-10-21  1:07 ` [PATCH v5 08/16] fuzz: add a DISABLE_PCI op to general-fuzzer Alexander Bulekov
2020-10-21  1:07 ` [PATCH v5 09/16] fuzz: add a crossover function to generic-fuzzer Alexander Bulekov
2020-10-21  1:07 ` Alexander Bulekov [this message]
2020-10-21  1:07 ` [PATCH v5 11/16] scripts/oss-fuzz: Add crash trace minimization script Alexander Bulekov
2020-10-21  1:07 ` [PATCH v5 12/16] fuzz: Add instructions for using general-fuzz Alexander Bulekov
2020-10-21  1:07 ` [PATCH v5 13/16] fuzz: add an "opaque" to the FuzzTarget struct Alexander Bulekov
2020-10-21  1:07 ` [PATCH v5 14/16] fuzz: add general-fuzz configs for oss-fuzz Alexander Bulekov
2020-10-21  1:07 ` [PATCH v5 15/16] fuzz: register predefined general-fuzz configs Alexander Bulekov
2020-10-21  1:07 ` [PATCH v5 16/16] scripts/oss-fuzz: remove the general-fuzz target Alexander Bulekov
2020-10-21 13:24 ` [PATCH v5 00/16] Add a Generic Virtual Device Fuzzer Thomas Huth

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20201021010752.973230-11-alxndr@bu.edu \
    --to=alxndr@bu.edu \
    --cc=bsd@redhat.com \
    --cc=darren.kenny@oracle.com \
    --cc=dimastep@yandex-team.ru \
    --cc=f4bug@amsat.org \
    --cc=pbonzini@redhat.com \
    --cc=qemu-devel@nongnu.org \
    --cc=stefanha@redhat.com \
    --cc=thuth@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.