ll documented under the
extension section. So at this point, we're still debating on how best
to place the text within the document for promotion to normative, while
at the same time also trying to adjust structured replies to make error
replies easier to identify.
--
Eric Blake ebla
Split into two parts, to make the value changes distinct from the
document reorganization
Eric Blake (2):
doc: Move sections about structured reply values
doc: In STRUCTURED_REPLY, make error types easy to recognize
doc/proto.md | 234
, tweak a couple of wording issues outside of the
section shuffling.
Signed-off-by: Eric Blake
---
doc/proto.md | 219 +--
1 file changed, 109 insertions(+), 110 deletions(-)
diff --git a/doc/proto.md b/doc/proto.md
index 29ef281..e484055
sending
something the client isn't expecting, but covering the
situation from both sides never hurts. Also, marking error
types resembles how NBD_REP_ERR_* has a common layout.
Signed-off-by: Eric Blake
---
doc/proto.md | 93 +++-
1 file ch
t's how it was before, and you seem to have reordered (not
> sure why).
Because of our earlier conversation on the same topic:
https://sourceforge.net/p/nbd/mailman/nbd-general/thread/20160409104438.GP19023%40grep.be/#msg35003527
--
Eric Blake eblake redhat com+1-919-301-3266
L
applies to the "magic 0x67446698" but NOT to the "bytes MAY
be invalid if an error", because there is no text above that states the
bytes may be invalid until the next paragraph).
--
Eric Blake eblake redhat com+1-919-301-3266
Libvirt virtualization library http://l
we think that it is unlikely enough to worry about, where by the time
there are more than 16 transmission flags, users are likely to already
be using new-enough qemu that doesn't send global flags to the kernel?
--
Eric Blake eblake redhat com+1-919-301-3266
Libvirt
On 04/21/2016 04:31 AM, Alex Bligh wrote:
>
> On 20 Apr 2016, at 16:48, Wouter Verhelst wrote:
>
>> On Mon, Apr 18, 2016 at 01:49:45PM -0600, Eric Blake wrote:
>>> plus simplifications to NBD_OPT_GO:
>>>
>>> diff --git i/doc/proto.md w/doc/proto.md
&
run by FSF) is
available, but it does a lot better at spam management. Also, qemu
recently aligned with the Software Freedom Conservancy
https://lists.gnu.org/archive/html/qemu-devel/2015-07/msg04802.html
and as such may have advice on how to migrate canonical management of a
project. Since qemu is one
Now that there is at most one in-flight message, we can drop
some other statements about the server needing to wait to
process one option while other options are in-flight.
Signed-off-by: Eric Blake
---
master branch
doc/proto.md | 4 +---
1 file changed, 1 insertion(+), 3 deletions(-)
diff
Now that we require at most one in-flight option request, we
can simplify the wording related to handling of NBD_OPT_GO
replies.
Signed-off-by: Eric Blake
---
extension-info branch
doc/proto.md | 8 +++-
1 file changed, 3 insertions(+), 5 deletions(-)
diff --git a/doc/proto.md b/doc
On 04/21/2016 09:49 PM, Alex Bligh wrote:
> Eric,
>
> On 21 Apr 2016, at 20:12, Eric Blake wrote:
>
>> Now that we require at most one in-flight option request, we
>> can simplify the wording related to handling of NBD_OPT_GO
>> replies.
>
> This l
utdown requests'
037/44:[down] 'nbd: Create struct for tracking export info'
038/44:[down] 'block: Add blk_get_opt_transfer_length()'
039/44:[0065] [FC] 'nbd: Implement NBD_OPT_GO on server'
040/44:[0139] [FC] 'nbd: Implement NBD_OPT_GO on client'
041/4
tion, just
> say that if it's provided inband that takes priority.
>
> Thoughts?
It seems reasonable to me. Should one of us propose a followup patch to
the extension-info branch with the simplification?
--
Eric Blake eblake redhat com+1-919-301-3266
Libvi
BLOCKSIZE stuff?
>
I don't think so, because the ability to read holes is the main reason I
added chunking of the read reply.
--
Eric Blake eblake redhat com+1-919-301-3266
Libvirt virtualization library http://libvirt.org
signature.asc
Description: OpenPGP digital signature
--
n the "Block Sizes" section as a justification for why a
server might want to advertise non-default sizes, if we get rid of the
mention here.
Otherwise, I like where this is headed in terms of simplification, and
it's only a few minor tweaks to my existing qemu patches to update to
the c
d on constraints on their values. The server MUST send this info if
> + it intends to enforce block sizes other than the defaults. After
> + sending this information, the server can legitimately assume that
> + any client that co
PORT in its request list (since that one is mandatory even
without being requested)? On the other hand, I don't see what it would
add other than verbosity.
Looking better, and I'm still feeling comfortable with the amount of
tweaks to my qemu patches to implement this change if we go wi
assumption cannot be made solely on the basis of an `NBD_OPT_INFO`
> + with an `NBD_INFO_BLOCK_SIZE` request).
s/request)/request, nor if the client did not request
`NBD_INFO_BLOCK_SIZE`)/
> The *length* MUST be 14,
> + and the reply payload is interpreted as:
>
>
n its next request for NBD_OPT_GO or to give up now.
--
Eric Blake eblake redhat com+1-919-301-3266
Libvirt virtualization library http://libvirt.org
signature.asc
Description: OpenPGP digital signature
--
Find and fi
igned-off-by: Eric Blake
---
extension-info branch
doc/proto.md | 13 +++--
1 file changed, 7 insertions(+), 6 deletions(-)
diff --git a/doc/proto.md b/doc/proto.md
index 1a0f1cf..da65df0 100644
--- a/doc/proto.md
+++ b/doc/proto.md
@@ -949,12 +949,13 @@ of the newstyle negoti
off (the server will happen to honor the request, rather than
disconnect or fail with the suggested EINVAL, but the client was
out-of-spec so it shouldn't be relying on any particular server
behavior, whether success or a particular error).
Let me know if any of the above reasoning is wrong, or
choice of preferred block size isn't as much of a sticking
point as the choice of the minimum block size. And a choice of whether
to use O_DIRECT might also be, to some extent, and out-of-band
configuration. At any rate, I'm glad we agree that my reading of the
spec vs. planned implementatio
On 03/31/2016 02:34 PM, Eric Blake wrote:
> On 03/31/2016 02:17 PM, Alex Bligh wrote:
>> OK so I actually went and researched what my answer was last time I
>> was asked ( :-) ):
>>
>> Here was my conclusion last time after trawling through lkml
>>
wn the
road. But not worth rewriting the commit message for that.
>
> Signed-off-by: Alex Bligh
> ---
> nbd-client.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
Reviewed-by: Eric Blake
>
> diff --git a/nbd-client.c b/nbd-client.c
> index cc9a06e..ddc849c 100
optional, but
if present they represent 8 spaces. Many GNU projects use a space-only
policy (GNU coreutils for example). So if I have any say, I'd recommend
that you re-tune your indent settings for space-only on any file that
you reindent to keep in the GNU style.
--
Eric Blak
the NBD protocol before
taking anything into qemu. We've got time on our hand, so let's use it
to get this right. (That, and I have several pending patches that
conflict with this as part of adding WRITE_ZEROES and INFO_BLOCK_SIZE
support, where it may be easier to resubmit this
qemu is free to do how it wishes.
Okay, sounds like that part is uncontroversial, and it is indeed worth
improving qemu's behavior.
>
> What surprises me is that a release kernel is using experimental
> NBD extensions; there's no guarantee these won't change. Or does
>
On 05/10/2016 09:41 AM, Alex Bligh wrote:
>
> On 10 May 2016, at 16:29, Eric Blake wrote:
>
>> So the kernel is currently one of the clients that does NOT honor block
>> sizes, and as such, servers should be prepared for ANY size up to
>> UINT_MAX (other than DoS
n("/dev/nbd1", O_RDWR) = 3
fallocate(3, FALLOC_FL_KEEP_SIZE|FALLOC_FL_PUNCH_HOLE, 0, 65536) = -1
ENODEV (No such device)
The ENODEV is a rather unusual choice of error.
--
Eric Blake eblake redhat com+1-919-301-3266
Libvirt virtualization library http://libvirt.org
si
series.
The cover letter is a great place to point out how v4 differs from v3,
but also to point out which patches are unchanged from v3, to save
reviewer's time. So if all you did was add new patches, a cover-letter
mention of which patches remain unchanged might be helpful.
--
Eric Blake
(if requested by client and replied by server,
then this can be larger than the normal maximum size in
NBD_INFO_BLOCK_SIZE; if either side doesn't request the info, then
assume any maximum in NBD_INFO_BLOCK_SIZE applies, otherwise UINT32_MAX;
and the two infos are separate items because
field (although then we'd have to
be careful about whether an NBD_REP message could have different
flags than those requested by the client).
Signed-off-by: Eric Blake
---
doc/proto.md | 28 +---
1 file changed, 25 insertions(+), 3 deletions(-)
diff --git a/doc/prot
message sizes (one for options, the other for
transmission).
--
Eric Blake eblake redhat com+1-919-301-3266
Libvirt virtualization library http://libvirt.org
signature.asc
Description: OpenPGP digital signature
---
only actually trim a subset
of the client request), CMD_WRITE_ZEROES must write actual zeroes
over the entire request (but if it can punch holes, only a
subset of the client request need result in holes).
Signed-off-by: Eric Blake
---
For the extension-info branch, and builds on the earlier
patch
On 05/11/2016 02:20 PM, Alex Bligh wrote:
>
> On 11 May 2016, at 20:08, Eric Blake wrote:
>
>> Calling out a MUCH smaller limit of 16k during handshake phase in
>> contrast to transmission phase, and acknowledging that we have bytes in
>> transmission phase that wi
to make
this interoperability point obvious.
>
> To test qemu against nbdkit you can do this in the nbdkit sources:
>
> make
> make check TESTS=test-newstyle \
> LIBGUESTFS_HV=/path/to/qemu/x86_64-softmmu/qemu-system-x86_64 \
>
ability point obvious.
>
> I can't actually see the issue here. It explains what needs to be
> implemented by the server, and that includes NBD_OPT_LIST. Very
> happy to add some clarity, but I'm not sure where it'
On 05/17/2016 09:52 AM, Eric Blake wrote:
>>> Perhaps nbdkit should implement NBD_OPT_LIST returning just "" (the
>>> default name) as its only list entry?
>>
>> Or "default".
>
> As I read the protocol, I don't see "default"
On 05/17/2016 09:58 AM, Richard W.M. Jones wrote:
> On Tue, May 17, 2016 at 09:52:30AM -0600, Eric Blake wrote:
>> so it might be nicer to
>> make a change to the protocol document that instead permits current
>> nbdkit behavior and puts the burden on clients to interoperate
n NBD_REP_SERVER with name "";
>> or an explicit statement that if a server rejects NBD_OPT_LIST, then the
>> client SHOULD assume that any name will work for
>> NBD_OPT_EXPORT_NAME/NBD_OPT_GO.
>
> I think it's sim
On 05/17/2016 10:41 AM, Richard W.M. Jones wrote:
> On Tue, May 17, 2016 at 10:05:50AM -0600, Eric Blake wrote:
>> On 05/17/2016 09:58 AM, Richard W.M. Jones wrote:
>>> On Tue, May 17, 2016 at 09:52:30AM -0600, Eric Blake wrote:
>>>> so it might be nicer to
>
On 05/17/2016 10:39 AM, Richard W.M. Jones wrote:
> On Tue, May 17, 2016 at 09:59:02AM -0600, Eric Blake wrote:
>> On 05/17/2016 09:52 AM, Eric Blake wrote:
>>>>> Perhaps nbdkit should implement NBD_OPT_LIST returning just "" (the
>>>>> default na
itrary name,
but even then, there is no guarantee that the qemu client is using the
"" default name, and since NBD_OPT_EXPORT_NAME doesn't care, I think
it's nicer to state that NBD_OPT_LIST should O
[adding nbd list]
On 06/13/2016 06:10 AM, Paolo Bonzini wrote:
>
>
> On 12/05/2016 00:39, Eric Blake wrote:
>> - If we report an error to NBD_CMD_READ, we are not writing out
>> any data payload; but the protocol says that a client can expect
>> to read the paylo
until the last user
> closes the block device. This offers the possibility as well to open a
> new socket to be used while the filesystems are mounted.
>
> Signed-off-by: Pranay Kr. Srivastava
>
> [mpa: Keep the blockdevice open until all users left]
> Signed-off-by: Markus Parg
; Signed-off-by: Pranay Kr. Srivastava
> ---
> drivers/block/nbd.c | 124
> ++++----
> 1 file changed, 96 insertions(+), 28 deletions(-)
>
--
Eric Blake eblake redhat com+1-919-301-3266
Libvirt virtuali
On 06/27/2016 06:13 AM, Paolo Bonzini wrote:
>
>
> On 26/06/2016 00:15, Eric Blake wrote:
>> diff --git a/block/nbd.c b/block/nbd.c
>> index 8d57220..049d1bd 100644
>> --- a/block/nbd.c
>> +++ b/block/nbd.c
>> @@ -357,6 +357,7 @@ static int nbd_co_flus
[adding nbd list]
On 07/19/2016 12:21 AM, Fam Zheng wrote:
> On Mon, 07/18 22:08, Eric Blake wrote:
>> Upstream NBD protocol recently added the ability to efficiently
>> write zeroes without having to send the zeroes over the wire,
>> along with a flag to control whether th
On 07/19/2016 09:34 PM, Fam Zheng wrote:
> On Tue, 07/19 17:45, Paolo Bonzini wrote:
>>
>>
>> On 19/07/2016 17:28, Eric Blake wrote:
>>>> If I'm reading the NBD proto.md correctly, this is not enough if
>>>> NBD_CMD_FLAG_NO_HOLE is specified.
On 09/15/2016 06:09 AM, Alex Bligh wrote:
>
> I also wonder whether any servers that can do caching per
> connection will always share a consistent cache between
> connections. The one I'm worried about in particular here
> is qemu-nbd - Eric Blake CC'd.
>
I doubt t
ation
>> stage (it would only be one flag, e.g. 'multiple connections
>> unsafe').
>
> I suppose that's not a bad idea.
Seems like it may need to be a per-export flag, rather than a global
flag (as a given server may be able to serve multiple types of files,
where t
ron out
the details of the documentation before making it a formal part of the
spec. It's important to remember that such a proposal should still be
optional (a server need not implement the new mode, and a client should
be prepared to fall back to other means if the server does not suppo
sly requested via NBD_CMD_FLAG_SHIFT, rather
than having ALL information it needs immediately available in the
current NBD_CMD_WRITE_ZEROES).
I'd much rather support a single flag that says to zero the entire disk
than to introduce stateful variable-amount shifting.
--
Eric Blake
ething that the client can change at-will during a live session,
may be useful. Or it may be the source of arithmetic overflow exploits
in poor implementations, or of denial-of-service with used with a READ
or WRITE to send more than 2G of data in a single command. In other
words, I don't y
vised to start using 'subdir-objects' option
throughout your
automake: project, to avoid future incompatibilities.
autoreconf: automake failed with exit status: 1
I ran out of time to try and patch it today, if you want to beat me to it.
--
Eric Blake eblake redhat com+1-919-301-3
}
send_reply(client, opt, NBD_REP_ERR_UNSUP, -1, "The
given option is
unknown to this server implementation");
break;
}
--
Eric Blake eblake redhat com+1-919-301-3266
Libvirt virtualization library http://libvirt.org
sig
ch as me
posting this email just as a heads-up in case anyone else hits the
problem (I just hit it in qemu while trying to add ESHUTDOWN support).
--
Eric Blake eblake redhat com+1-919-301-3266
Libvirt virtualizat
ging away at the autoreconf failure on my end. This was
just the quick PoC to get the conversation going, pointing out that you
made a common mistake (since I've already fixed the same bug in both
qemu and nbdkit); and that it is the same mistake that
NBD_FLAG_FIXED_NEWSTYLE is supposed to adver
On 10/14/2016 01:32 PM, Wouter Verhelst wrote:
> On Thu, Oct 13, 2016 at 05:33:18PM -0500, Eric Blake wrote:
>> I'm getting this failure when trying to build NBD, as part of
>> 'autoreconf -vfi':
>>
>> configure.ac:248: error: required file 'system
On 10/14/2016 02:23 PM, Eric Blake wrote:
>> static void socket_read(CLIENT* client, void *buf, size_t len) {
>> +void *tmp = NULL;
>> +if (!buf) {
>> +/* FIXME: Enforce maximum bound on client-provided len? */
>> +
omake: of the corresponding sources.
automake: You are advised to start using 'subdir-objects' option throughout your
automake: project, to avoid future incompatibilities.
Following the advice doesn't break anything, so let's do it.
I also noticed that the build leaves behind an
y for so long is that it has
taken us this long to finally want to implement clients that use
a new option.
Signed-off-by: Eric Blake
---
Perhaps you want to split this into two patches, one adding
consume_len() and fixing the default case that responds with
NBD_REP_ERR_UNSUP (with inte
On 10/14/2016 04:02 PM, Eric Blake wrote:
> Any client attempting to probe support for a new option, such as
> NBD_OPT_STARTTLS or NBD_OPT_GO, with plans to do a graceful
> fallback to older methods, will fail in its attempt if the server
> does not ignore the length field and potentia
On 10/14/2016 04:02 PM, Eric Blake wrote:
> /**
> * Consume data from a socket that we don't want
> *
> - * @param f a file descriptor
> + * @param c the client data stream
> * @param buf a buffer
> * @param len the number of bytes to consume
> * @param
=/path/to/qemutls --tls-creds tls0 --exportname default
where /path/to/qemutls contains the necessary certificate files.
--
Eric Blake eblake redhat com+1-919-301-3266
Libvirt virtualization library http://libvirt.org
signature.asc
Description: OpenPGP digital signature
--
On 10/15/2016 01:46 PM, Wouter Verhelst wrote:
> On Fri, Oct 14, 2016 at 03:11:55PM -0500, Eric Blake wrote:
>> On 10/14/2016 01:32 PM, Wouter Verhelst wrote:
>>> On Thu, Oct 13, 2016 at 05:33:18PM -0500, Eric Blake wrote:
>>>> I'm getting this failure w
On 10/14/2016 03:19 PM, Eric Blake wrote:
> Starting from a fresh git checkout, running ./autogen.sh gives a
> couple of warnings on my Fedora 24 build tools, one from libtool:
>
> libtoolize: Consider adding '-I support' to ACLOCAL_AMFLAGS in Makefile.am.
>
> and o
v2 of a couple of patches I've posted in the last week, now
rolled into one series, fixing a bug triggered by automake 1.15,
and adding testsuite coverage (which in turn exposed another
recent buggy commit).
Eric Blake (6):
build: Silence autogen.sh warnings
server: Fix botched s
ients.
Furthermore, even if the server requires TLS, and rejects all but
NBD_OPT_STARTTLS as the first valid option, it should still honor
NBD_OPT_ABORT.
Signed-off-by: Eric Blake
---
nbd-server.c | 8
1 file changed, 8 insertions(+)
diff --git a/nbd-server.c b/nbd-server.c
index 4b0692d..d5b2013 1
ff-by: Eric Blake
---
nbd-server.c | 16
1 file changed, 16 insertions(+)
diff --git a/nbd-server.c b/nbd-server.c
index c93a9d8..4b0692d 100644
--- a/nbd-server.c
+++ b/nbd-server.c
@@ -373,6 +373,21 @@ static inline void consume(CLIENT* c, size_t len, void *
buf, size_t b
Commit 3b80382 tried to make it easy for the server to send an
error message whose length was determined by strlen(), but ended
up sending a length of UINT32_MAX, causing clients to either
hang up (reply too large) or wait for nearly 4G of data that was
never coming.
Signed-off-by: Eric Blake
nstead open-code the traversal to the desired files.
I also noticed that the build was already leaving behind an untracked
manpage.log file, in addition to the new .dirstamp witness file created
by our new use of subdir-objects.
Signed-off-by: Eric Blake
---
.gitignore| 2 ++
Makefile.
The signature of consume() threw me off. Good design says that if
you are going to have paired parameters (buf and bufsize), you
generally want them adjacent, not separated by an unrelated parameter
(len). Move len to be first, adjusting all callers.
Signed-off-by: Eric Blake
---
nbd-server.c
unknown command.
Signed-off-by: Eric Blake
---
tests/run/nbd-tester-client.c | 47 +++
1 file changed, 47 insertions(+)
diff --git a/tests/run/nbd-tester-client.c b/tests/run/nbd-tester-client.c
index 28db8ee..099bec1 100644
--- a/tests/run/nbd-tester-client.c
ry-picked two commits from Josef
> to allow building on CentOS 6, would be silly if we lose that again...)
And that will be part of my testing.
--
Eric Blake eblake redhat com+1-919-301-3266
Libvirt virtualization
On 10/18/2016 09:12 AM, Eric Blake wrote:
> On 10/18/2016 06:01 AM, Wouter Verhelst wrote:
>
>>> Following the advice almost works, except that automake 1.15 still
>>> has a nasty bug (https://debbugs.gnu.org/cgi/bugreport.cgi?bug=13928)
>>> where use of $(foo) in
o allow building on CentOS 6, would be silly if we lose that again...)
I'll make sure it works before sending v3 of this patch. However, the
other patches in this series should still be usable as-is without this
one going in.
--
Eric Blake eblake redhat com+1-919-301-3266
Libvirt v
new use of subdir-objects.
This patch has been tested with 'make distcheck' across multiple
automake and libtool versions, ranging from CentOS 6 vintage to current
git toolchains.
Signed-off-by: Eric Blake
---
.gitignore| 2 ++
Makefile.am | 1 +
configure.ac
These two patches replace 1/6 in my previous posting, while
2-6 of that series remain identical (and can even be applied
independently prior to this respin, other than the review
comment on 5/6 that still needs addressing).
Eric Blake (2):
build: Distribute netdb-compat.h without relying on
nbd-server depends on netdb-compat.h; however, we were only
including it in the tarball as a side effect of it also being
used by the testsuite. Make the dependency explicit.
Signed-off-by: Eric Blake
---
Makefile.am | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a
On 10/18/2016 06:07 AM, Wouter Verhelst wrote:
> On Mon, Oct 17, 2016 at 03:23:39PM -0500, Eric Blake wrote:
>> Add a test of intentionally provoking a server error during option
>> negotiation to prove that a client can still fall back to known
>> options; thus covering two r
dd it within a .dir-locals.el file to make it apply to fresh git
checkouts?
--
Eric Blake eblake redhat com+1-919-301-3266
Libvirt virtualization library http://libvirt.org
signature.asc
Description: OpenPGP digital sign
er sending the wrong length in an error reply, and
for a server not reading enough data when replying to an unknown
command.
Signed-off-by: Eric Blake
---
As requested by Wouter, split this off into an independent test,
rather than impacting the startup of all other tests.
tests/run/M
On 10/20/2016 02:38 AM, Wouter Verhelst wrote:
> On Wed, Oct 19, 2016 at 05:07:11PM -0500, Eric Blake wrote:
>> I noticed that nbd source code prefers the use of hard tab instead of
>> spaces for indentation, and emacs does not play nicely with that by
>> default. Does anyone
On 10/20/2016 03:55 AM, Alex Bligh wrote:
>
>> On 19 Oct 2016, at 23:07, Eric Blake wrote:
>>
>> I noticed that nbd source code prefers the use of hard tab instead of
>> spaces for indentation, and emacs does not play nicely with that by
>> default.
-off-by: Eric Blake
---
.dir-locals.el | 2 ++
1 file changed, 2 insertions(+)
create mode 100644 .dir-locals.el
diff --git a/.dir-locals.el b/.dir-locals.el
new file mode 100644
index 000..046e004
--- /dev/null
+++ b/.dir-locals.el
@@ -0,0 +1,2 @@
+((c-mode . ((c-file-style . "
h) to it? That enum was meant as that "framework" ;-)
>
my v3 changed the TEST_ enum instead, but I'll see if I can redo it to
trigger off of a CONNECTION_TYPE_ value.
--
Eric Blake eblake redhat com+1-919-301-3266
Libvirt virtualization libr
Commit b885246 creates a symlink to work around an automake weakness,
but forgot to ignore the link when doing an in-tree build.
Signed-off-by: Eric Blake
---
Hmm. I guess I was so busy testing VPATH builds that I forgot to
do one final test of an in-tree build :)
.gitignore | 1 +
1 file
Signed-off-by: Josef Bacik
> ---
> drivers/block/nbd.c | 23 ---
> 1 file changed, 12 insertions(+), 11 deletions(-)
>
--
Eric Blake eblake redhat com+1-919-301-3266
Libvirt virtualization library http://libvirt.org
signature.asc
Descri
While not directly related to NBD_CMD_WRITE_ZEROES, the qemu
team discovered that it is useful if a server can advertise
whether an export is in a known-all-zeroes state at the time
the client connects.
Signed-off-by: Eric Blake
---
doc/proto.md | 5 +
1 file changed, 5 insertions(+)
This
On 12/06/2016 03:25 AM, Kevin Wolf wrote:
> Am 06.12.2016 um 00:42 hat Eric Blake geschrieben:
>> While not directly related to NBD_CMD_WRITE_ZEROES, the qemu
>> team discovered that it is useful if a server can advertise
>> whether an export is in a known-all-zeroes sta
ly tell the server to advertise the bit because the
>> user has side knowledge that the file was just created (and then the
>> burden of misbehavior is on the user if they mistakenly request the
>> advertisement when it is not true).
>
> Maybe that's the only practical app
;> clarifies what a client should do with that information if it gets it
>> (i.e., "don't read it, it doesn't contain anything interesting").
>
> That's fair enough until the last bit in brackets. Rather than saying
> a client SHOULD NOT read it, it should s
read as
> +all zeroes; if clear, the block contents are not known. Note
> +that the use of `NBD_CMD_WRITE_ZEROES` is related to this
> +status, but that the server MAY report zeroes even where write
> +zeroes has not been requ
e inter-branch diff is indeed the best way to review the
current state of the changes in relation to the master branch, rather
than trying to follow one patch at a time. I'm grateful that you've
stepped in to try and nail down some of the wordings, and doing a qemu
proof-of-concept implementation
in the (16-bit) "transmission flags" field, where space is
> at a premium.
No, it is NOT a "transmission flag", as it is a per-export property
(where we currently have 64 bits).
--
Eric Blake e
On 12/14/2016 02:22 AM, Wouter Verhelst wrote:
> Hi Eric,
>
> On Tue, Dec 13, 2016 at 04:36:08PM -0600, Eric Blake wrote:
>> On 12/13/2016 06:18 AM, Wouter Verhelst wrote:
>>> On Tue, Dec 13, 2016 at 08:38:12AM +0100, Kevin Wolf wrote:
>>>> Am 12.12.2016 um 1
other than colons,
> +and be non-empty. The entire name (namespace, colon and leaf-name)
> +MUST NOT exceed 255 bytes (and therefore botht he namespace and
s/botht he/both the/
--
Eric Blake eblake redhat com+1-919-301-3266
Libvirt virtualization library http://libvi
er. Admittedly
> I didn't have a 'real' client to test it against.
It's working well in qemu 2.8 without needing tweaks to the
documentation. Should we try and do some cross-implementation testing
today, before doing the actual merge?
--
Eric Blake eblake redhat com
201 - 300 of 338 matches
Mail list logo
