Author: ken
Date: Wed Jan 23 03:05:19 2019
New Revision: 21021

Log:
Rust: I was mistaken about the Ryzen problem with the testsuite, traps and a 
few segfaults are normal but nobody usually looks at the syslog.

Modified:
   trunk/BOOK/general/prog/rust.xml
   trunk/BOOK/introduction/welcome/changelog.xml

Modified: trunk/BOOK/general/prog/rust.xml
==============================================================================
--- trunk/BOOK/general/prog/rust.xml    Wed Jan 23 01:11:41 2019        (r21020)
+++ trunk/BOOK/general/prog/rust.xml    Wed Jan 23 03:05:19 2019        (r21021)
@@ -201,25 +201,17 @@
       by a message that the build completed successfully.
     </para>
 
-    <caution>
+    <note>
       <para>
-        On AMD Ryzen processors (family 17h), the non-optimized version of 
libstd
-        which is compiled at the start of the tests contains one or more 
opcodes
-        which are not implemented on this CPU family. These will be logged in 
the
+        The testsuite will generate some messages in the
         <phrase revision="sysv">system log</phrase>
         <phrase revision="systemd">systemd journal</phrase>
-        and will be followed a few minutes later by segmentation faults. 
Despite
-        that, the tests continue to run, apparently normally. But the system 
may
-        reboot before the tests have completed. The normal optimized libraries
-        run without this problem.
+        for traps on invalid opcodes, and for segmentation faults.
+        In themselves these are nothing to worry about, although if the
+        output from the testsuite reports tests which FAIL with such faults
+        then there may be a problem.
       </para>
-      <para>
-        A mitigation is to install gdb
-        <!-- systemd apparently handles this with systemd-coredump -->
-        <phrase revision="sysv">and to run the tests with 'ulimit -C 
disabled'</phrase>
-        but this does not always prevent the system rebooting.
-      </para>
-    </caution>
+    </note>
 
     <para>
       To run the tests issue

Modified: trunk/BOOK/introduction/welcome/changelog.xml
==============================================================================
--- trunk/BOOK/introduction/welcome/changelog.xml       Wed Jan 23 01:11:41 
2019        (r21020)
+++ trunk/BOOK/introduction/welcome/changelog.xml       Wed Jan 23 03:05:19 
2019        (r21021)
@@ -45,6 +45,15 @@
       <para>January 23rd, 2018</para>
       <itemizedlist>
         <listitem>
+          <para>[ken] - Change the 'Caution' on the rustc page (about running
+          the testsuite on AMD Ryzens) to a 'Note', the reboot problem was
+          almost certainly kernel triple-faults on an -rc kernel, although
+          the testsuite seemed to reliably trigger it at that time. The
+          supposed mitigation (now removed) was clearly irrelevant, and to
+          top it all the invalid opcodes also appear on other machines, so
+          they too are probably deliberate.</para>
+        </listitem>
+        <listitem>
           <para>[pierre] - Update dependencies for libtasn1 and GTK+3:
           the dependency on python-six is obsolete.</para>
         </listitem>
-- 
http://lists.linuxfromscratch.org/listinfo/blfs-book
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Reply via email to