Code Review - #arch/interrupts

2 views
Skip to first unread message

Davide Libenzi

unread,
Dec 3, 2015, 3:57:08 PM12/3/15
to Akaros
This is unfortunately on top of devarch_msr_perf (the perf one posted yesterday), since I needed per-CPU, which lives there.


BTW, there seem to be something wrong with block extra data.


The following changes since commit 87782eff25975f0d52c3778ce93b0c496f653f83:

  Added new perf utility to access CPU counters from userspace (2015-12-02 16:13:50 -0800)

are available in the git repository at:

  g...@github.com:dlibenzi/akaros interrupt_device

for you to fetch changes up to 23b44ab22a56704642d79f665cbca9cbc6b727e1:

  Added devarch interrupts device (2015-12-03 12:46:38 -0800)

----------------------------------------------------------------
Davide Libenzi (3):
      Added sizeof_field() macro
      Allows the queue data structure to use a minimum block size
      Added devarch interrupts device

 kern/arch/riscv/devarch.h |   8 +++
 kern/arch/x86/devarch.c   | 125 ++++++++++++++++++++++++++++++++++++++++++++++
 kern/arch/x86/devarch.h   |   1 +
 kern/arch/x86/trap.c      |   3 ++
 kern/include/ns.h         |   4 +-
 kern/include/ros/common.h |   2 +
 kern/src/ns/qio.c         |  71 ++++++++++++++++++++++----
 7 files changed, 203 insertions(+), 11 deletions(-)
 create mode 100644 kern/arch/riscv/devarch.h

Davide Libenzi

unread,
Dec 7, 2015, 6:41:29 PM12/7/15
to Akaros
The qpread code has been dropped, and the dual ended buffer has been added.

$ cat \#arch/interrupts
           IRQ[ 49]  IRQ[240]
CPU[  0]         55      1533
CPU[  1]          0         0
CPU[  2]          0         0
CPU[  3]          0         0
CPU[  4]          0         0
CPU[  5]          0         0
CPU[  6]          0         0
CPU[  7]          0         0

Davide Libenzi

unread,
Dec 11, 2015, 3:12:57 PM12/11/15
to Akaros
This has been rebased with the perf branch as well, upon which it was based (per CPU requirements).

Davide Libenzi

unread,
Dec 14, 2015, 5:37:20 PM12/14/15
to Akaros
One more rebase due to the latest changes in the perf branch.

Davide Libenzi

unread,
Dec 15, 2015, 3:31:16 PM12/15/15
to Akaros
And one more rebase ...

Davide Libenzi

unread,
Dec 21, 2015, 1:52:02 PM12/21/15
to Akaros
Reply all
Reply to author
Forward
0 new messages