Bug#779926: pu: package intel-microcode/1.20150121.1
Package: release.debian.org
Severity: normal
User: release.debian.org@packages.debian.org
Usertags: pu
I'd like to update the intel-microcode package in Wheezy to the latest
available public Intel microcode.
This intel-microcode release (20150121) has been tested in unstable and
also Debian jessie for one month, without any error reports.
It updates only the microcode for Intel Desktop/mobile Broadwell E0/F0
processors, such as the Core i7-5500U and a few others.
It also updates the initramfs scripts, to decouple the intel-microcode
and amd64-microcode packages. I will need this for a future stable
update of amd64-microcode. These changes have been tested for oven one
year in Debian unstable and Debian jessie, without issues.
I've attached the abridged debdiff, without the upstream microcode
changes.
diffstat:
changelog | 11
debian/changelog | 47
debian/initramfs.hook | 2
debian/initramfs.init-premount | 19
microcode-20140913.dat |40694 ----------------------------------------
microcode-20150121.dat |41591 +++++++++++++++++++++++++++++++++++++++++
6 files changed, 41663 insertions(+), 40701 deletions(-)
Thank you.
diff -Nru intel-microcode-1.20140913.1/changelog intel-microcode-1.20150121.1/changelog
--- intel-microcode-1.20140913.1/changelog 2014-10-30 16:14:19.000000000 -0200
+++ intel-microcode-1.20150121.1/changelog 2015-02-11 20:32:44.000000000 -0200
@@ -1,3 +1,14 @@
+2015-01-21:
+ * Downgraded microcodes (to a previously shipped revision):
+ sig 0x000306f2, pf mask 0x6f, 2014-09-03, rev 0x0029, size 28672
+
+2015-01-07:
+ * New Microcodes:
+ sig 0x000306d4, pf mask 0xc0, 2014-12-05, rev 0x0018, size 14336
+
+ * Updated Microcodes (this update is known to cause issues):
+ sig 0x000306f2, pf mask 0x6f, 2014-11-21, rev 0x002d, size 28672
+
2014-09-13:
* New Microcodes:
sig 0x000306f2, pf mask 0x6f, 2014-09-03, rev 0x0029, size 28672
diff -Nru intel-microcode-1.20140913.1/debian/changelog intel-microcode-1.20150121.1/debian/changelog
--- intel-microcode-1.20140913.1/debian/changelog 2014-12-18 16:31:28.000000000 -0200
+++ intel-microcode-1.20150121.1/debian/changelog 2015-03-01 23:33:19.000000000 -0300
@@ -1,3 +1,50 @@
+intel-microcode (1.20150121.1) stable; urgency=high
+
+ * New upstream microcode data file 20150121
+ + Downgraded microcodes (to a previously shipped revision):
+ sig 0x000306f2, pf mask 0x6f, 2014-09-03, rev 0x0029, size 28672
+ * The microcode downgrade fixes a very nasty regression on Xeon E5v3
+ processors (closes: #776431)
+ * critical urgency: the broken sig 0x306f2, rev 0x2b microcode shipped
+ in release 20150107 caused CPU core hangs and Linux boot failures.
+ The upstream fix was to downgrade it to the same microcode revision
+ that was shipped in release 20140913
+ * source: remove superseded upstream data file: 20150107.
+
+ -- Henrique de Moraes Holschuh <hmh@debian.org> Fri, 30 Jan 2015 08:41:20 -0200
+
+intel-microcode (1.20150107.1) stable; urgency=high
+
+ * New upstream microcode data file 20150107
+ + New Microcodes:
+ sig 0x000306d4, pf mask 0xc0, 2014-12-05, rev 0x0018, size 14336
+ + Updated Microcodes:
+ sig 0x000306f2, pf mask 0x6f, 2014-11-21, rev 0x002d, size 28672
+ + High urgency: there are fast-tracked microcode updates in this
+ release which imply that critical errata are being fixed
+ (Broadwell Core i3/i5/i7 5th gen, Core M-5Y, Pentium 3805U,
+ Celeron 3755U, maybe others)
+ * source: remove superseded upstream data file: 20140913
+ * initramfs: decouple from amd64-microcode:
+ Update the initramfs init-premount boot script to the script used in
+ intel-microcode 1.20130222.6 to 1.20130808.2, as well as all
+ intel-microcode 2.x packages. It has been throughoutly tested for
+ more than one year in unstable, testing (jessie), and
+ wheezy-backports. This new version of the boot script decouples
+ intel-microcode from amd64-microcode's boot script, and will trigger
+ a microcode update only when an Intel processor is installed.
+ amd64-microcode's boot script runs earlier, so this change will at
+ most cause a microcode update to be triggered twice (the kernel will
+ ignore the second attempt). Therefore, it is compatible with any
+ version of the amd64-microcode package. This change allows
+ amd64-microcode's boot script to also be updated to decouple itself
+ from intel-microcode.
+ * initramfs.hook: do not mix arrays and lists.
+ Avoid echo "foo $@", use echo "foo $*" instead. This is unlikely
+ to be expĺoitable, but it makes ShellCheck happier.
+
+ -- Henrique de Moraes Holschuh <hmh@debian.org> Sun, 18 Jan 2015 19:17:01 -0200
+
intel-microcode (1.20140913.1) stable; urgency=low
* New upstream microcode data file 20140913
diff -Nru intel-microcode-1.20140913.1/debian/initramfs.hook intel-microcode-1.20150121.1/debian/initramfs.hook
--- intel-microcode-1.20140913.1/debian/initramfs.hook 2014-12-18 16:31:28.000000000 -0200
+++ intel-microcode-1.20150121.1/debian/initramfs.hook 2015-03-01 23:33:19.000000000 -0300
@@ -27,7 +27,7 @@
verbose()
{
if [ "${verbose}" = "y" ] ; then
- echo "intel-microcode: $@"
+ echo "intel-microcode: $*"
fi
:
}
diff -Nru intel-microcode-1.20140913.1/debian/initramfs.init-premount intel-microcode-1.20150121.1/debian/initramfs.init-premount
--- intel-microcode-1.20140913.1/debian/initramfs.init-premount 2014-12-18 16:31:28.000000000 -0200
+++ intel-microcode-1.20150121.1/debian/initramfs.init-premount 2015-03-01 23:33:19.000000000 -0300
@@ -23,15 +23,22 @@
;;
esac
-# Only continue if we do possibly have firmware to install
-[ -d /lib/firmware/intel-ucode -o -d /lib/firmware/amd-ucode ] || exit 0
+# hack to strip spaces
+is_intel()
+{
+ [ "x$1" = "xvendor_id" ] && [ "x$2" = "xGenuineIntel" ] && HAS_INTL=1
+}
-# we don't need to run twice
-[ -e ${0%/*}/amd64-microcode ] && exit 0
-[ -e ${0%/*}/amd64_microcode ] && exit 0
+# Only continue if we do possibly have firmware to install
+[ -d /lib/firmware/intel-ucode ] || exit 0
# module load will already have caused the microcode to be fetched
-{ while read -r module trash ; do [ "x$module" = "xmicrocode" ] && exit 0 ; done } < /proc/modules
+{ while read -r module trash ; do [ "x$module" = "xmicrocode" ] && exit 0 ; done ; } < /proc/modules
+
+# don't run on non-Intel
+HAS_INTL=0
+{ while IFS=: read -r field value ; do is_intel $field $value ; done ; } < /proc/cpuinfo
+[ $HAS_INTL -eq 0 ] && exit 0;
. /scripts/functions
diff -Nru intel-microcode-1.20140913.1/microcode-20140913.dat intel-microcode-1.20150121.1/microcode-20140913.dat
diff -Nru intel-microcode-1.20140913.1/microcode-20150121.dat intel-microcode-1.20150121.1/microcode-20150121.dat
Reply to: