Add histogram to track DNS outcomes after malformed responses

The special error handling for this case is strange and I want to delete
it. This histogram will track how often it's useful so we can decide
whether it's okay to delete.

Bug: 779589
Change-Id: Ieadcb1214486bd468e5e9c18d2f06915b6ee1fff
Reviewed-on: https://chromium-review.googlesource.com/822832Reviewed-by: Julia Tuttle <juliatuttle@chromium.org>
Reviewed-by: Steven Holte <holte@chromium.org>
Commit-Queue: Miriam Gershenson <mgersh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#524188}

net/dns/dns_transaction.cc

@@ -71,6 +71,26 @@ std::unique_ptr<base::Value> NetLogStartCallback(
   return std::move(dict);
 }
 
+// Values are used in UMA histograms. Do not change existing values.
+enum MalformedResponseResult {
+  MALFORMED_OK = 0,
+  MALFORMED_MALFORMED = 1,
+  MALFORMED_FAILED = 2,
+  MALFORMED_MAX
+};
+
+void RecordMalformedResponseHistogram(int net_error) {
+  MalformedResponseResult error_type;
+  if (net_error == OK)
+    error_type = MALFORMED_OK;
+  else if (net_error == ERR_DNS_MALFORMED_RESPONSE)
+    error_type = MALFORMED_MALFORMED;
+  else
+    error_type = MALFORMED_FAILED;
+  UMA_HISTOGRAM_ENUMERATION("Net.DNS.ResultAfterMalformedResponse", error_type,
+                            MALFORMED_MAX);
+}
+
 // ----------------------------------------------------------------------------
 
 // A single asynchronous DNS exchange, which consists of sending out a
@@ -205,10 +225,15 @@ class DnsUDPAttempt : public DnsAttempt {
     } while (rv != ERR_IO_PENDING && next_state_ != STATE_NONE);
 
     set_result(rv);
-    // If we received a malformed response, and are now waiting for another one,
-    // indicate to the transaction that the server might be misbehaving.
-    if (rv == ERR_IO_PENDING && received_malformed_response_)
-      return ERR_DNS_MALFORMED_RESPONSE;
+    if (received_malformed_response_) {
+      // If we received a malformed response, and are now waiting for another
+      // one, indicate to the transaction that the server might be misbehaving.
+      if (rv == ERR_IO_PENDING)
+        return ERR_DNS_MALFORMED_RESPONSE;
+
+      // This is a new response after the original malformed one.
+      RecordMalformedResponseHistogram(rv);
+    }
     if (rv == OK) {
       DCHECK_EQ(STATE_NONE, next_state_);
       UMA_HISTOGRAM_LONG_TIMES_100("AsyncDNS.UDPAttemptSuccess",
@@ -263,6 +288,7 @@ class DnsUDPAttempt : public DnsAttempt {
       // Our solution is to make another attempt, in case the query truly
       // failed, but keep this attempt alive, in case it was a false alarm.
       received_malformed_response_ = true;
+      RecordMalformedResponseHistogram(ERR_DNS_MALFORMED_RESPONSE);
       next_state_ = STATE_READ_RESPONSE;
       return OK;
     }

tools/metrics/histograms/enums.xml

@@ -26465,6 +26465,12 @@ from previous Chrome versions.
   </int>
 </enum>
 
+<enum name="MalformedResponseResult">
+  <int value="0" label="Success"/>
+  <int value="1" label="Malformed"/>
+  <int value="2" label="Other failure"/>
+</enum>
+
 <enum name="ManagedUserPasswordChange">
   <int value="0" label="OK_MANAGER">Changed in manager session</int>
   <int value="1" label="OK_MANGED">Changed in supervised user session</int>

tools/metrics/histograms/histograms.xml

@@ -42379,6 +42379,17 @@ http://cs/file:chrome/histograms.xml - but prefer this file for new entries.
   </summary>
 </histogram>
 
+<histogram name="Net.DNS.ResultAfterMalformedResponse"
+    enum="MalformedResponseResult">
+  <owner>mgersh@chromium.org</owner>
+  <summary>
+    When DnsUDPAttempt gets a malformed response, it returns an error but
+    continues running. This histogram records all attempt outcomes that are
+    returned to the transaction once a malformed response has been received,
+    including the original malformed response.
+  </summary>
+</histogram>
+
 <histogram name="Net.DNS.TotalTime" units="ms">
   <owner>mgersh@chromium.org</owner>
   <summary>