Skip to content

Commit fc47992

Browse files
committed
skiboot 5.1.19 release notes
Signed-off-by: Stewart Smith <[email protected]>
1 parent 41cace5 commit fc47992

File tree

1 file changed

+46
-0
lines changed

1 file changed

+46
-0
lines changed

doc/release-notes/skiboot-5.1.19.txt

+46
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,46 @@
1+
skiboot-5.1.19
2+
--------------
3+
4+
skiboot-5.1.19 was released on Monday 16th January 2017.
5+
6+
skiboot-5.1.19 is the 20th stable release of 5.1, it follows skiboot-5.1.18
7+
(which was released 26th August 2016).
8+
9+
This release contains a few minor bug fixes.
10+
11+
Changes are:
12+
13+
Generic:
14+
- Makefile: Disable stack protector due to gcc problems
15+
- stack: Don't recurse into __stack_chk_fail
16+
- Makefile: Use -ffixed-r13
17+
- we did not find evidence of this ever being a problem, but this fix
18+
is good and preventative.
19+
- Limit number of "Poller recursion detected" errors to display
20+
In some error conditions, we could spiral out of control on this
21+
and spend all of our time printing the exact same backtrace.
22+
Limit it to 16 times, because 16 is a nice number.
23+
24+
FSP based Systems:
25+
- fsp: Don't recurse pollers in ibm_fsp_terminate
26+
If we were to terminate in a poller, we'd call op_display() which
27+
called pollers which hit the recursive poller warning, which ended
28+
in not much fun at all.
29+
30+
PCI:
31+
- hw/phb3: set PHB retry state correctly when fresetting during a creset
32+
- phb3: Lock the PHB on set_xive callbacks
33+
Those are called by the interrupts core and thus skip the locking
34+
implicit in the PCI opal calls.
35+
- hw/{phb3, p7ioc}: Return success for freset on empty PHB
36+
OPAL_CLOSED is returned when fundamental reset is issued on the
37+
PHB who doesn't have subordinate devices (root port excluded).
38+
The kernel raises an error message, which is unnecessary. This
39+
returns OPAL_SUCCESS for this case to avoid the error message.
40+
- hw/phb3: fix error handling in complete reset
41+
During a complete reset, when we get a timeout waiting for pending
42+
transaction in state PHB3_STATE_CRESET_WAIT_CQ, we mark the PHB as broken
43+
and return OPAL_PARAMETER.
44+
Change the return code to OPAL_HARDWARE which is way more sensible, and set
45+
the state to PHB3_STATE_FENCED so that the kernel can retry the complete
46+
reset.

0 commit comments

Comments
 (0)