1. make deliver_intr a linux irq handler, and do not let it touch the
main_intr_queue. the queue
handling is solely in intr_thread (removing dead entries) and
insert_intr_entry (insertion).
2. remove user_intr from struct linux_action and its handling on linux side.
---
device/intr.c
1. The queue manipulated by two threads, the registration server and the
intr_thread, so we need a lock to protect it.
2. To initialize the lock, we introduce a irq_init function in
i386/i386/irq.[ch] and call it in device/device_init.c
---
device/device_init.c | 3 ++-
device/intr.c|
Ok. Will do.
Junling
> On Aug 2, 2020, at 4:52 PM, Samuel Thibault wrote:
>
> Junling Ma, le dim. 02 août 2020 16:45:22 -0700, a ecrit:
>> I guess you misunderstood what I mean. I mean that there seems to be no
>> interest in the discussion on IRC yesterday that we introduce a new scheme.
>
>
Junling Ma, le dim. 02 août 2020 16:45:22 -0700, a ecrit:
> I guess you misunderstood what I mean. I mean that there seems to be no
> interest in the discussion on IRC yesterday that we introduce a new scheme.
That's because you asked on IRC at a random time.
That's what IRC is not good at: coll
I guess you misunderstood what I mean. I mean that there seems to be no
interest in the discussion on IRC yesterday that we introduce a new scheme. If
so, I will not work on that part. For part I, ie., moving the irq management to
mach side, it people think it is useful, I will prepare a series
Samuel Thibault, le lun. 03 août 2020 01:33:53 +0200, a ecrit:
> Junling Ma, le dim. 02 août 2020 16:19:39 -0700, a ecrit:
> > You may have missed the part where I say let user side [/dev/irq*]
> > translator
> > do the multiplexing.
>
> I didn't but that looks like extra complexity, and makes th
Junling Ma, le dim. 02 août 2020 16:19:39 -0700, a ecrit:
> I was distracted by using a counter at all in the original code,
> because an irq cannot be fired more than once before it is acked.
Right, normally that doesn't happen.
> What we really need is a boolean_t type per notification,
And th
> On Aug 2, 2020, at 3:44 PM, Samuel Thibault wrote:
>
> Junling Ma, le dim. 02 août 2020 15:32:01 -0700, a ecrit:
1. Change devicer_user_intr to be a linux interrupt handler.
>>>
>>> Right, this makes sense, but did you actually try it? Check the existing
>>> code twice to realize why it'
Junling Ma, le dim. 02 août 2020 15:32:01 -0700, a ecrit:
> >> 1. Change devicer_user_intr to be a linux interrupt handler.
> >
> > Right, this makes sense, but did you actually try it? Check the existing
> > code twice to realize why it's done that way. Here linux_intr is looking
> > through the
Hi Samual,
Thanks for reviewing.
> Junling Ma, le sam. 01 août 2020 18:59:58 -0700, a ecrit:
>> PATCH 1 is included in this email: it prepares the stage by
>
> Please provide separate patches for each over these. Otherwise this is
> all conflated and way more difficult to review.
Ok.
>> 1. Ch
I applied, thanks!
Just one thing: I moved the functions so as to get the code in the
execution order: fsys_startup, then fsys_init, then
arrange_shutdown_notification, and S_startup_dosync before
trivfs_goaway.
Samuel
Damien Zammit, le sam. 01 août 2020 15:05:38 +1000, a ecrit:
> +static void
> +device_shutdown (void)
> +{
> + struct block_data *bd = block_head;
> +
> + while (bd)
> +{
> + device_close((void *)bd);
> + bd = bd->next;
> +}
> + rump_sys_reboot (0, NULL);
> +}
Did you check th
Samuel Thibault, le dim. 02 août 2020 23:16:22 +0200, a ecrit:
> Damien Zammit, le sam. 01 août 2020 15:09:57 +1000, a ecrit:
> > I figured out why it was hanging on reboot:
> > netdde was frozen because pci-arbiter was not working
> > with my setup that used x86 method of pci access for rumpdisk.
Hello,
Damien Zammit, le sam. 01 août 2020 15:09:57 +1000, a ecrit:
> I figured out why it was hanging on reboot:
> netdde was frozen because pci-arbiter was not working
> with my setup that used x86 method of pci access for rumpdisk.
>
> When I removed the translators from /dev/netdde and /serve
Could you also send a patch that fixes libddekit? Otherwise, again, I
can't apply your patches.
Until I would happen to magically find the time to fix libddekit myself
of course.
Samuel
Junling Ma, le sam. 01 août 2020 19:06:02 -0700, a ecrit:
> In this patch, the old interface of device_intr_register and device_intr_ack
> is removed.
We don't want to remove it yet, we need to keep it for smooth upgrade.
Please everybody always remember to test smooth upgrades. Otherwise I
end
Junling Ma, le sam. 01 août 2020 19:04:05 -0700, a ecrit:
> In this patch, the new interface of irq device is implemented. Please see the
> first patch for a description.
Not only, it also brings other changes, making the patch unreviewable.
Samuel
Hello,
Junling Ma, le sam. 01 août 2020 18:59:58 -0700, a ecrit:
> PATCH 1 is included in this email: it prepares the stage by
Please provide separate patches for each over these. Otherwise this is
all conflated and way more difficult to review.
> 1. Change devicer_user_intr to be a linux interr
18 matches
Mail list logo
