On Tue, Sep 06, 2022 at 04:19:40AM -0500, Eric Blake wrote:
> > > + assert (cr3 = (Int64.add cr2 fudge))
>
> Not so in OCaml, where + isn't even polymorphic to int64, so I have to
> break out manual calls to Int64.XXX methods. Here, I went with a
> helper variable 'fudge'.
No implicit type conv
On Mon, Sep 05, 2022 at 01:25:27PM +0200, Laszlo Ersek wrote:
> This concern arises because in the child process, only the one thread
> returning from fork() exists, thus if any locks are held by other threads
> in the parent process, those will never be released in the child process.
>
> But such
On Mon, Sep 05, 2022 at 01:25:28PM +0200, Laszlo Ersek wrote:
> Each call to table_attach() in fact open-codes (row, row + 1), for setting
> the top and bottom grid lines of the widget being added to the table.
> What's more, within a given table, we add widgets in a left-to-right
> first, top-down
On Mon, Sep 05, 2022 at 01:25:29PM +0200, Laszlo Ersek wrote:
> Introduce a checkbox that lets the user choose between options (a) and
> (b), as outlined in one of the previous patches in this series. When the
> checkbox is ticked -- that is, the fully populated physical topology is
> copied --, ma
On Mon, Sep 05, 2022 at 01:25:27PM +0200, Laszlo Ersek wrote:
> + "p2v.vcpu.dense_topo" => manual_entry->new(
> +shortopt => "", # ignored for booleans
> +description => "
> +Copy the physical machine's CPU topology, densely populated, to the
> +guest. Disabled by default. If disabled, t
On Mon, Sep 05, 2022 at 01:25:31PM +0200, Laszlo Ersek wrote:
> This lets us exercise both states of the "p2v.vcpu.dense_topo" switch
> sensibly via the in-VM GUI.
>
> Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1590721
> Signed-off-by: Laszlo Ersek
> ---
> Makefile.am | 2 ++
> 1 file
On 09/06/22 11:38, Eric Blake wrote:
> On Mon, Sep 05, 2022 at 04:35:28PM +0200, Laszlo Ersek wrote:
>> Now, in case I'm wrong, and we enter this code with a *third*
>> "h->opt_current" value, then we have:
>>
>> - opt = SET, and
>>
>> - "h->opt_current" differing from both SET and LIST.
>>
>> The
On 09/06/22 11:08, Eric Blake wrote:
> On Mon, Sep 05, 2022 at 02:00:22PM +0200, Laszlo Ersek wrote:
> We could further compress things to have:
>
> pr "debug (h, \"leave: ret=\"";
> (match ret with
> | RBool -> pr "%%d\", ret"
>
> and so on, to get rid of yet another spurious pair of "" in
On 09/06/22 11:19, Eric Blake wrote:
> On Mon, Sep 05, 2022 at 02:10:36PM +0200, Laszlo Ersek wrote:
>> On 09/03/22 00:14, Eric Blake wrote:
>>> This proves that the stat counters increment as desired, as well as
>>> proving that our RUInt32 generator type works.
>>
>> How is this related to RUint3
On 09/08/22 10:03, Richard W.M. Jones wrote:
> On Mon, Sep 05, 2022 at 01:25:27PM +0200, Laszlo Ersek wrote:
>> + "p2v.vcpu.dense_topo" => manual_entry->new(
>> +shortopt => "", # ignored for booleans
>> +description => "
>> +Copy the physical machine's CPU topology, densely populated, to
On 09/08/22 09:57, Richard W.M. Jones wrote:
> On Mon, Sep 05, 2022 at 01:25:28PM +0200, Laszlo Ersek wrote:
>> Each call to table_attach() in fact open-codes (row, row + 1), for setting
>> the top and bottom grid lines of the widget being added to the table.
>> What's more, within a given table, w
On 09/08/22 10:04, Richard W.M. Jones wrote:
> On Mon, Sep 05, 2022 at 01:25:31PM +0200, Laszlo Ersek wrote:
>> This lets us exercise both states of the "p2v.vcpu.dense_topo" switch
>> sensibly via the in-VM GUI.
>>
>> Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1590721
>> Signed-off-by:
(Adding Dan for input)
On Thu, Sep 08, 2022 at 03:23:41PM +0200, Laszlo Ersek wrote:
> On 09/08/22 10:03, Richard W.M. Jones wrote:
> > On Mon, Sep 05, 2022 at 01:25:27PM +0200, Laszlo Ersek wrote:
> >> + "p2v.vcpu.dense_topo" => manual_entry->new(
> >> +shortopt => "", # ignored for booleans
On 09/08/22 15:36, Richard W.M. Jones wrote:
> (Adding Dan for input)
>
> On Thu, Sep 08, 2022 at 03:23:41PM +0200, Laszlo Ersek wrote:
>> On 09/08/22 10:03, Richard W.M. Jones wrote:
>>> On Mon, Sep 05, 2022 at 01:25:27PM +0200, Laszlo Ersek wrote:
+ "p2v.vcpu.dense_topo" => manual_entry->n
On Mon, Sep 05, 2022 at 01:25:27PM +0200, Laszlo Ersek wrote:
> Currently, the CPU topology for the converted domain is determined as
> follows:
>
> (1) main() [main.c]
>
> (2) set_config_defaults() [main.c]
> vcpus <-- sy
On Thu, Sep 08, 2022 at 02:36:15PM +0100, Richard W.M. Jones wrote:
> (Adding Dan for input)
>
> On Thu, Sep 08, 2022 at 03:23:41PM +0200, Laszlo Ersek wrote:
> > On 09/08/22 10:03, Richard W.M. Jones wrote:
> > > On Mon, Sep 05, 2022 at 01:25:27PM +0200, Laszlo Ersek wrote:
> > >> + "p2v.vcpu.de
On 09/08/22 16:37, Daniel P. Berrangé wrote:
> On Mon, Sep 05, 2022 at 01:25:27PM +0200, Laszlo Ersek wrote:
>> Currently, the CPU topology for the converted domain is determined as
>> follows:
>>
>> (1) main() [main.c]
>>
>> (2) set_config_defaults()
On 09/08/22 16:39, Daniel P. Berrangé wrote:
> On Thu, Sep 08, 2022 at 02:36:15PM +0100, Richard W.M. Jones wrote:
>> (Adding Dan for input)
>>
>> On Thu, Sep 08, 2022 at 03:23:41PM +0200, Laszlo Ersek wrote:
>>> On 09/08/22 10:03, Richard W.M. Jones wrote:
On Mon, Sep 05, 2022 at 01:25:27PM +
Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1590721
v1:
https://listman.redhat.com/archives/libguestfs/2022-September/029806.html
Please see the Notes section on each patch for the updates in this
version (addressing v1 feedback). Patch #5 is a candidate for dropping
in particular
Introduce a checkbox that lets the user choose between options (a) and
(b), as outlined in one of the previous patches in this series. When the
checkbox is ticked -- that is, the fully populated physical topology is
copied --, make the VCPU count text entry insensitive, and overwrite its
contents w
show_conversion_dialog() hides the VCPU and memory limit warning signs
unconditionally. The warning signs and the associated label text(s) can
only be displayed later, when the VCPUs or the memory entry changes, and
vcpus_or_memory_check_callback() is called.
This is incorrect: we may initialize e
Each call to table_attach() in fact open-codes (row, row + 1), for setting
the top and bottom grid lines of the widget being added to the table.
What's more, within a given table, we add widgets in a left-to-right
first, top-down second fashion, so "row" only ever increases within a
particular tabl
This lets us exercise both states of the "p2v.vcpu.phys_topo" switch
sensibly via the in-VM GUI.
Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1590721
Signed-off-by: Laszlo Ersek
Acked-by: Richard W.M. Jones
---
Notes:
v2:
- pick up Rich's A-b
- s/dense_topo/phys_t
Currently, the CPU topology for the converted domain is determined as
follows:
(1) main() [main.c]
(2) set_config_defaults() [main.c]
vcpus <-- sysconf (_SC_NPROCESSORS_ONLN)
get_cpu_config()
Such a default seems reasonable for most phys-to-virt conversions.
Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1590721
Signed-off-by: Laszlo Ersek
Acked-by: Richard W.M. Jones
---
Notes:
v2:
- pick up Rich's A-b
- resolve rebase conflicts due to dense_topo -> phy
On Thu, Sep 08, 2022 at 02:53:21PM +0200, Laszlo Ersek wrote:
> On 09/06/22 11:08, Eric Blake wrote:
> > On Mon, Sep 05, 2022 at 02:00:22PM +0200, Laszlo Ersek wrote:
>
> > We could further compress things to have:
> >
> > pr "debug (h, \"leave: ret=\"";
> > (match ret with
> > | RBool -> pr
On 08/31/22 16:39, Eric Blake wrote:
> As promised in the previous patch, also test the new
> nbd_set_request_meta_context() API in Python, OCaml, and Golang. No
> direct C counterpart in tests/, but it's always good to check that our
> language bindings are complete.
> ---
> python/t/110-default
On 08/31/22 16:39, Eric Blake wrote:
> When listing information about an export, we do not use
> nbd_block_status() or nbd_can_meta_context(), because we were already
> utilizing the callback to nbd_opt_list_meta_context() to collect
> supported context names. In practice, because we also never ca
On 09/01/22 17:08, Eric Blake wrote:
> On Wed, Aug 31, 2022 at 09:39:23AM -0500, Eric Blake wrote:
>> Since NBD_OPT_LIST_META_CONTEXTS is stateless from the server's point
>> of view, there is no reason to make it clear state on the client's
>> side. Since we do not otherwise modify h->meta_contex
On 08/31/22 16:39, Eric Blake wrote:
> Add test coverage for the previous patch not wiping state during
> nbd_opt_list_meta_context. The change is logically identical in each
> language that has the same unit test.
> ---
> python/t/240-opt-list-meta.py | 29 +-
> ocaml/tests/tes
On 08/31/22 16:39, Eric Blake wrote:
> The documentation for nbd_internal_reset_size_and_flags() claims we
> should wipe all knowledge of a previously-negotiated export when
> swapping export names. However, we weren't actually doing that, which
> could lead to a user calling nbd_opt_info() for on
31 matches
Mail list logo
