I am leaving IBM and will move on to other working area,
so remove myself as a vfio-ccw maintainer.
Signed-off-by: Dong Jia Shi
---
I've enjoyed working with you guys. Hope to meet you in future,
until then, thank you for all your coach and help, please take
care, and farewell.
[I'v
t was freed.
Suggested-by: Pierre Morel ,
Signed-off-by: Dong Jia Shi
---
drivers/s390/cio/vfio_ccw_cp.c | 82 +++---
1 file changed, 36 insertions(+), 46 deletions(-)
diff --git a/drivers/s390/cio/vfio_ccw_cp.c b/drivers/s390/cio/vfio_ccw_cp.c
index b38ef387e4f0..2
Fixed typos.
Added sanity check on pa->pa_iova_pfn and updated comments accordingly.
- #4. Removed unused idaw_nr.
- #5. Replaced leading white spaces with TABs.
Traced the function in anycase.
Dong Jia Shi (3):
vfio: ccw: shorten kernel doc description for pfn_array_pin()
vfio:
From: Halil Pasic
Add some tracepoints so we can inspect what is not working as is should.
Signed-off-by: Halil Pasic
Signed-off-by: Dong Jia Shi
---
drivers/s390/cio/Makefile | 1 +
drivers/s390/cio/vfio_ccw_fsm.c | 17 +++-
drivers/s390/cio/vfio_ccw_trace.h | 54
The kernel doc description for usage of the struct pfn_array in
pfn_array_pin() is unnecessary long. Let's shorten it by describing
the contents of the struct pfn_array fields at the struct's definition
instead.
Suggested-by: Cornelia Huck
Signed-off-by: Dong Jia Shi
---
driver
Let's avoid free on ccw->cda that points to a guest address
or an already freed memory area by setting it to NULL if memory
allocation didn't happen or failed.
Signed-off-by: Dong Jia Shi
---
drivers/s390/cio/vfio_ccw_cp.c | 31 ---
1 file changed,
From: Halil Pasic
Add some tracepoints so we can inspect what is not working as is should.
Signed-off-by: Halil Pasic
Signed-off-by: Dong Jia Shi
---
drivers/s390/cio/Makefile | 1 +
drivers/s390/cio/vfio_ccw_fsm.c | 17 +++-
drivers/s390/cio/vfio_ccw_trace.h | 54
From: Dong Jia Shi
This refactors pfn_array_alloc_pin() and also improves it by adding
defensive code in error handling so that calling pfn_array_unpin_free()
after error return won't lead to problem. This mainly does:
1. Merge pfn_array_pin() into pfn_array_alloc_pin(), since there
From: Dong Jia Shi
Let's avoid free on ccw->cda that points to a guest address
or an already freed memory area by setting it to NULL if memory
allocation didn't happen or failed.
Signed-off-by: Dong Jia Shi
---
drivers/s390/cio/vfio_ccw_cp.c | 31 --
From: Dong Jia Shi
The kernel doc description for usage of the struct pfn_array in
pfn_array_pin() is unnecessary long. Let's shorten it by describing
the contents of the struct pfn_array fields at the struct's definition
instead.
Suggested-by: Cornelia Huck
Signed-off-by: Do
ts accordingly.
- #4. Removed unused idaw_nr.
- #5. Replaced leading white spaces with TABs.
Traced the function in anycase.
Dong Jia Shi (3):
vfio: ccw: shorten kernel doc description for pfn_array_pin()
vfio: ccw: refactor and improve pfn_array_alloc_pin()
vfio: ccw: set cc
If device has not been registered, or there is a pending work,
we should reschedule a sch_event call again.
Signed-off-by: Dong Jia Shi
---
drivers/s390/cio/vfio_ccw_drv.c | 5 -
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/drivers/s390/cio/vfio_ccw_drv.c b/drivers/s390/cio
elements are excluded) cleanup activities
based on false assumptions can be avoided.
Let's make sure cp_free works properly after cp_prefetch returns with an
error by setting ch_len of a ccw chain to the number of the translated
CCWs on that chain.
Acked-by: Pierre Morel
Reviewed-by: Dong Ji
Let's avoid free on ccw->cda that points to a guest address
or an already freed memory area by setting it to NULL if memory
allocation didn't happen or failed.
Signed-off-by: Dong Jia Shi
---
drivers/s390/cio/vfio_ccw_cp.c | 31 ---
1 file changed,
t was freed.
Suggested-by: Pierre Morel ,
Signed-off-by: Dong Jia Shi
---
drivers/s390/cio/vfio_ccw_cp.c | 82 +++---
1 file changed, 36 insertions(+), 46 deletions(-)
diff --git a/drivers/s390/cio/vfio_ccw_cp.c b/drivers/s390/cio/vfio_ccw_cp.c
index e8fe7450702e..3
The kernel doc description for usage of the struct pfn_array in
pfn_array_pin() is unnecessary long. Let's shorten it by describing
the contents of the struct pfn_array fields at the struct's definition
instead.
Suggested-by: Cornelia Huck
Signed-off-by: Dong Jia Shi
---
driver
From: Halil Pasic
Add some tracepoints so we can inspect what is not working as is should.
Signed-off-by: Halil Pasic
Signed-off-by: Dong Jia Shi
---
drivers/s390/cio/Makefile | 1 +
drivers/s390/cio/vfio_ccw_fsm.c | 16 +++-
drivers/s390/cio/vfio_ccw_trace.h | 77
po fixes.
- #2. New patch.
- #3. Added the missing suggested-by Pierre.
Fixed typos.
Added sanity check on pa->pa_iova_pfn and updated comments accordingly.
- #4. Removed unused idaw_nr.
- #5. Replaced leading white spaces with TABs.
Traced the function in anycase.
Dong
elements are excluded) cleanup activities
based on false assumptions can be avoided.
Let's make sure cp_free works properly after cp_prefetch returns with an
error by setting ch_len to 0 for the ccw chains those are not prefetched.
Acked-by: Pierre Morel
Reviewed-by: Dong Jia Shi
Signed-o
Let's avoid free on ccw->cda that points to a guest address
or a already freed memory area by setting it to NULL if memory
allocation didn't happen or failed.
Signed-off-by: Dong Jia Shi
---
drivers/s390/cio/vfio_ccw_cp.c | 28 +++-
1 file changed, 19 ins
From: Halil Pasic
Add some tracepoints so we can inspect what is not working as is should.
Signed-off-by: Halil Pasic
Signed-off-by: Dong Jia Shi
---
drivers/s390/cio/Makefile | 1 +
drivers/s390/cio/vfio_ccw_fsm.c | 13 ++
drivers/s390/cio/vfio_ccw_trace.h | 86
er of pfn_array_pin(). As a result, also remove kernel-doc
for pfn_array_pin() and add kernel-doc for pfn_array_alloc_pin().
2. For a vfio_pin_pages() failure, set pa->pa_nr to zero to indicate
zero pages were pinned.
3. Set pa->pa_iova_pfn to NULL right after it was freed.
Signed-off-by: D
idual patches.
Thanks.
Dong Jia Shi (2):
vfio: ccw: refactor and improve pfn_array_alloc_pin()
vfio: ccw: set ccw->cda to NULL defensively
Halil Pasic (2):
vfio: ccw: fix cleanup if cp_prefetch fails
vfio: ccw: add traceponits for interesting error paths
drivers/s390/cio/Makefile
t?
>
> I'm not sure about the later. What prevents the guest from believing
> the (to use your terminology shared) chp 0.00 has become unavailable
> if 0.00 becomes unavailable in the host?
>
> Would that adversely affect the virtual devices? It would at
> least look strange.
AFAI read the code, this series does not hurt the vritual devices, which
always have fixed path information in SCHIB.
The strange thing is, the chp 0.00 is both virtual and non-virtual with
mix use of virtual and non-virtual devices. And any existing problem is
not caused or enhanced by this work. If we can not fix the problem
without MCSS support or chp re-modelling in the near future, we should
document the problem and the effect.
I will do some test on this.
>
>>
>> We'd have to document this either I think.
>>
>>> Note: this is another unpleasant effect of not having MCSSE in the
>>> guest. The original design was to have a separate css for vfio, and
>>> then we would not have this kind of a problem. (Virtualization of
>>> chps would still be good for migration.)
>> Nod. BTW, I don't say that I do not want channel path virtualization in
>> the long run. It's just I want a minimal function set more currently
>> from what I'm focusing perspective.
>>
>> THANKS for these insightful questions!
>>
--
Dong Jia Shi
up being ignored)?
>> You mean the clash that sharing path 0.00 between a physical dasd and
>> the virtio ccw devices? The problem I saw is in the checking of the
>> chpid.is_virtual in css_do_rchp(). For a virtual chp, we should generate
>> INIT CRWs; while for a non-virtual one, we shouldn't. If the chp is
>> shared with virtual and non-virtual device, I don't know what to do
>> then...
>>
>> I don't think we can fix this before we re-modelled the channel path.
>> But this is neither introduced nor aggravated by this series, right?
>
> I'm not sure about the later. What prevents the guest from believing
> the (to use your terminology shared) chp 0.00 has become unavailable
> if 0.00 becomes unavailable in the host?
>
> Would that adversely affect the virtual devices? It would at
> least look strange.
>
>>
>> We'd have to document this either I think.
>>
>>> Note: this is another unpleasant effect of not having MCSSE in the
>>> guest. The original design was to have a separate css for vfio, and
>>> then we would not have this kind of a problem. (Virtualization of
>>> chps would still be good for migration.)
>> Nod. BTW, I don't say that I do not want channel path virtualization in
>> the long run. It's just I want a minimal function set more currently
>> from what I'm focusing perspective.
>>
>> THANKS for these insightful questions!
>>
--
Dong Jia Shi
This introduces a new irq for vfio-ccw to provide channel path
related event for userland.
Signed-off-by: Dong Jia Shi
---
drivers/s390/cio/vfio_ccw_ops.c | 29 +
drivers/s390/cio/vfio_ccw_private.h | 2 ++
include/uapi/linux/vfio.h | 1 +
3 files
the schib likeness.
Signed-off-by: Dong Jia Shi
---
hw/vfio/ccw.c | 24 +++-
1 file changed, 23 insertions(+), 1 deletion(-)
diff --git a/hw/vfio/ccw.c b/hw/vfio/ccw.c
index 16713f2c52..e673695509 100644
--- a/hw/vfio/ccw.c
+++ b/hw/vfio/ccw.c
@@ -32,6 +32,10 @@ typedef
This is a placeholder for a linux-headers update.
Signed-off-by: Dong Jia Shi
---
linux-headers/linux/vfio.h | 2 ++
linux-headers/linux/vfio_ccw.h | 6 ++
2 files changed, 8 insertions(+)
diff --git a/linux-headers/linux/vfio.h b/linux-headers/linux/vfio.h
index 4312e961ff..24d85be022
path information
out once got the eventfd signal, and do further process.
Signed-off-by: Dong Jia Shi
---
hw/vfio/ccw.c | 92 +--
1 file changed, 70 insertions(+), 22 deletions(-)
diff --git a/hw/vfio/ccw.c b/hw/vfio/ccw.c
index e673695509
le we are at it, touch one line of the comment around by
adding white space to make it aligned.
Signed-off-by: Dong Jia Shi
---
hw/s390x/css.c | 83 -
hw/s390x/s390-ccw.c | 20 +++
hw/vfio/ccw.c
path information update, we only do update if we were signaled.
We sets scsw.pno and pmcw.pnom to indicate path related event
for a guest. This is a bit lazy, but it works.
Signed-off-by: Dong Jia Shi
---
hw/s390x/css.c | 14 +--
hw/s390x/s390-ccw.c | 12 +
hw
Hi Folks,
This is the QEMU couterpart for the "basic channel path event handling" series.
For more information, please refer to the kernel counterpart.
Dong Jia Shi (5):
vfio: linux-headers update for vfio-ccw
vfio/ccw: get schib region info
vfio/ccw: get irq info and set up h
This introduces a new region for vfio-ccw to provide subchannel
information for user space.
Signed-off-by: Dong Jia Shi
---
drivers/s390/cio/vfio_ccw_fsm.c | 21 ++
drivers/s390/cio/vfio_ccw_ops.c | 79 +++--
drivers/s390/cio/vfio_ccw_private.h
This adds channel path related event handler for vfio-ccw.
This also signals userland when there is a chp event.
Signed-off-by: Dong Jia Shi
---
drivers/s390/cio/vfio_ccw_drv.c | 51 +
drivers/s390/cio/vfio_ccw_fsm.c | 22
drivers
Use PIM PAM POM CHPIDs
--
0.0.3f3f 0.0.0002 3390/0c 3990/e9 f0 f0 ff 42434445
#Notice: PAM changed again.
Dong Jia Shi (3):
vfio: ccw: introduce schib region
vfio: ccw: introduce channel path irq
vfio: ccw: handle chp event
drivers/s390/cio/
If the count field of a ccw is zero, there is no need to
try to pin page(s) for it. Let's check the count value
before starting pinning operations.
Reviewed-by: Pierre Morel
Signed-off-by: Dong Jia Shi
---
drivers/s390/cio/vfio_ccw_cp.c | 18 +-
1 file changed, 17 inser
ahead of
the bad idaw) could possibly be executed by the device before
I/O conclusion.
Suggested-by: Halil Pasic
Reviewed-by: Pierre Morel
Signed-off-by: Dong Jia Shi
---
drivers/s390/cio/vfio_ccw_cp.c | 4
1 file changed, 4 deletions(-)
diff --git a/drivers/s390/cio/vfio_ccw_cp.c b/dr
Some bugfixes according to recent discussion.
Dong Jia Shi (2):
vfio: ccw: bypass bad idaw address when fetching IDAL ccws
vfio: ccw: validate the count field of a ccw before pinning
drivers/s390/cio/vfio_ccw_cp.c | 22 +-
1 file changed, 17 insertions(+), 5 deletions
38 matches
Mail list logo
