> -----Original Message-----
> From: Klaus Jensen <its@irrelevant.dk>
> Sent: Monday, September 28, 2020 2:37 AM
> To: Dmitry Fomichev <Dmitry.Fomichev@wdc.com>
> Cc: Keith Busch <kbusch@kernel.org>; Damien Le Moal
> <Damien.LeMoal@wdc.com>; Klaus Jensen <k.jensen@samsung.com>; Kevin
> Wolf <kwolf@redhat.com>; Philippe Mathieu-Daudé <philmd@redhat.com>;
> Maxim Levitsky <mlevitsk@redhat.com>; Fam Zheng <fam@euphon.net>;
> Niklas Cassel <Niklas.Cassel@wdc.com>; qemu-block@nongnu.org; qemu-
> devel@nongnu.org; Alistair Francis <Alistair.Francis@wdc.com>; Matias
> Bjorling <Matias.Bjorling@wdc.com>
> Subject: Re: [PATCH v4 00/14] hw/block/nvme: Support Namespace Types
> and Zoned Namespace Command Set
> 
> On Sep 28 02:33, Dmitry Fomichev wrote:
> > > -----Original Message-----
> > > From: Klaus Jensen <its@irrelevant.dk>
> > >
> > > If it really needs to be memory mapped, then I think a hostmem-based
> > > approach similar to what Andrzej did for PMR is needed (I think that
> > > will get rid of the CONFIG_POSIX ifdef at least, but still leave it
> > > slightly tricky to get it to work on all platforms AFAIK).
> >
> > Ok, it looks that using the HostMemoryBackendFile backend will be
> > more appropriate. This will remove the need for conditional compile.
> >
> > The mmap() portability is pretty decent across software platforms.
> > Any poor Windows user who is forced to emulate ZNS on mingw will be
> > able to do so, just without having zone state persistency. Considering
> > how specialized this stuff is in first place, I estimate the number of users
> > affected by this "limitation" to be exactly zero.
> >
> 
> QEMU is a cross platform project - we should strive for portability.
> 
> Alienating developers that use a Windows platform and calling them out
> as "poor" is not exactly good for the zoned ecosystem.
> 

Wow. By bringing up political correctness here you are basically admitting
the fact that you have no real technical argument here. The whole Windows
issue is red herring that you are using to attack the code that is absolutely
legit, but comes from a competitor. Your initial complaint was that it
doesn't compile in mingw and that it uses "wrong" API. You have even
suggested the API to use. Now, the code uses that API and builds fine, but
now it's still not good simply because you "do not like it". It's a disgrace.

> > > But really,
> > > since we do not require memory semantics for this, then I think the
> > > abstraction is fundamentally wrong.
> > >
> >
> > Seriously, what is wrong with using mmap :) ? It is used successfully for
> > similar applications, for example -
> > https://github.com/open-iscsi/tcmu-runner/blob/master/file_zbc.c
> >
> 
> There is nothing fundamentally wrong with mmap. I just think it is the
> wrong abstraction here (and it limits portability for no good reason).
> For PMR there is a good reason - it requires memory semantics.
> 

We are trying to emulate NVMEe controller NVRAM.  The best abstraction
for emulating NVRAM would be... NVRAM!

> > > I am, of course, blowing my own horn, since my implementation uses a
> > > portable blockdev for this.
> > >
> >
> > You are making it sound like the entire WDC series relies on this approach.
> > Actually, the persistency is introduced in the second to last patch in the
> > series and it only adds a couple of lines of code in the i/o path to mark
> > zones dirty. This is possible because of using mmap() and I find the way
> > it is done to be quite elegant, not ugly :)
> >
> 
> No, I understand that your implementation works fine without
> persistance, but persistance is key. That is why my series adds it in
> the first patch. Without persistence it is just a toy. And the QEMU
> device is not just an "NVMe-version" of null_blk.
> 
> And I don't think I ever called the use of mmap ugly. I called out the
> physical memory API shenanigans as a hack.
> 
> > > Another issue is the complete lack of endian conversions. Does it
> > > matter? It depends. Will anyone ever use this on a big endian host and
> > > move the meta data backing file to a little endian host? Probably not.
> > > So does it really matter? Probably not, but it is cutting corners.
> > >
> 
> After I had replied this, I considered a follow-up, because there are
> probably QEMU developers that would call me out on this.
> 
> This definitely DOES matter to QEMU.
> 
> >
> > Great point on endianness! Naturally, all file backed values are stored in
> > their native endianness. This way, there is no extra overhead on big endian
> > hardware architectures. Portability concerns can be easily addressed by
> > storing metadata endianness as a byte flag in its header. Then, during
> > initialization, the metadata validation code can detect the possible
> > discrepancy in endianness and automatically convert the metadata to the
> > endianness of the host. This part is out of scope of this series, but I would
> > be able to contribute such a solution as an enhancement in the future.
> >
> 
> It is not out of scope. I don't see why we should merge something that
> is arguably buggy.

Again, wow! Now you turned around and arbitrarily elevated this issue from
moderate ("Does it matter?, cutting corners") to severe ("buggy"). Likely
because v5 of WDC patchset has been posted. This, again, just shows your
lack of integrity as a maintainer.

This "issue" is a real trivial one to fix as I described above and you are
blowing it up way out of proportion, making it look like it is a fundamental
problem that can not be resolved. It's not.

> 
> Bottomline is that I just don't see why we should accept an
> implementation that
> 
>   a) excludes some platforms (Windows) from using persistence; and
>   b) contains endianness conversion issues
> 
> when there is a portable implementation posted that at least tries to
> convert endianness as needed.

Doesn't that implementation discriminate against big endian architectures? :)
Ok, it is a joke - with some folks I need to clarify this.

On Sep 29 15:42, Dmitry Fomichev wrote:
> > -----Original Message-----
> > From: Klaus Jensen <its@irrelevant.dk>
> > Sent: Monday, September 28, 2020 2:37 AM
> > To: Dmitry Fomichev <Dmitry.Fomichev@wdc.com>
> > Cc: Keith Busch <kbusch@kernel.org>; Damien Le Moal
> > <Damien.LeMoal@wdc.com>; Klaus Jensen <k.jensen@samsung.com>; Kevin
> > Wolf <kwolf@redhat.com>; Philippe Mathieu-Daudé <philmd@redhat.com>;
> > Maxim Levitsky <mlevitsk@redhat.com>; Fam Zheng <fam@euphon.net>;
> > Niklas Cassel <Niklas.Cassel@wdc.com>; qemu-block@nongnu.org; qemu-
> > devel@nongnu.org; Alistair Francis <Alistair.Francis@wdc.com>; Matias
> > Bjorling <Matias.Bjorling@wdc.com>
> > Subject: Re: [PATCH v4 00/14] hw/block/nvme: Support Namespace Types
> > and Zoned Namespace Command Set
> > 
> > On Sep 28 02:33, Dmitry Fomichev wrote:
> > > > -----Original Message-----
> > > > From: Klaus Jensen <its@irrelevant.dk>
> > > >
> > > > If it really needs to be memory mapped, then I think a hostmem-based
> > > > approach similar to what Andrzej did for PMR is needed (I think that
> > > > will get rid of the CONFIG_POSIX ifdef at least, but still leave it
> > > > slightly tricky to get it to work on all platforms AFAIK).
> > >
> > > Ok, it looks that using the HostMemoryBackendFile backend will be
> > > more appropriate. This will remove the need for conditional compile.
> > >
> > > The mmap() portability is pretty decent across software platforms.
> > > Any poor Windows user who is forced to emulate ZNS on mingw will be
> > > able to do so, just without having zone state persistency. Considering
> > > how specialized this stuff is in first place, I estimate the number of users
> > > affected by this "limitation" to be exactly zero.
> > >
> > 
> > QEMU is a cross platform project - we should strive for portability.
> > 
> > Alienating developers that use a Windows platform and calling them out
> > as "poor" is not exactly good for the zoned ecosystem.
> > 
> 
> Wow. By bringing up political correctness here you are basically admitting
> the fact that you have no real technical argument here.

I prefer that we support all platforms if and when we can. That's a
technical argument, not a personal one like you those you start using
now.

> The whole Windows issue is red herring that you are using to attack
> the code that is absolutely legit, but comes from a competitor.

I can't even...

> Your initial complaint was that it doesn't compile in mingw and that
> it uses "wrong" API. You have even suggested the API to use. Now, the
> code uses that API and builds fine, but now it's still not good simply
> because you "do not like it". It's a disgrace.
> 

I answered this in a previous reply.

> > > > But really,
> > > > since we do not require memory semantics for this, then I think the
> > > > abstraction is fundamentally wrong.
> > > >
> > >
> > > Seriously, what is wrong with using mmap :) ? It is used successfully for
> > > similar applications, for example -
> > > https://github.com/open-iscsi/tcmu-runner/blob/master/file_zbc.c
> > >
> > 
> > There is nothing fundamentally wrong with mmap. I just think it is the
> > wrong abstraction here (and it limits portability for no good reason).
> > For PMR there is a good reason - it requires memory semantics.
> > 
> 
> We are trying to emulate NVMEe controller NVRAM.  The best abstraction
> for emulating NVRAM would be... NVRAM!
> 

You never brought that up before and sure it could be a fair argument,
except it is not true.

PMR is emulating NVRAM (and requires memory semantics). Persistent state
is not emulating anything. It is an implementation detail.

> > > > I am, of course, blowing my own horn, since my implementation uses a
> > > > portable blockdev for this.
> > > >
> > >
> > > You are making it sound like the entire WDC series relies on this approach.
> > > Actually, the persistency is introduced in the second to last patch in the
> > > series and it only adds a couple of lines of code in the i/o path to mark
> > > zones dirty. This is possible because of using mmap() and I find the way
> > > it is done to be quite elegant, not ugly :)
> > >
> > 
> > No, I understand that your implementation works fine without
> > persistance, but persistance is key. That is why my series adds it in
> > the first patch. Without persistence it is just a toy. And the QEMU
> > device is not just an "NVMe-version" of null_blk.
> > 
> > And I don't think I ever called the use of mmap ugly. I called out the
> > physical memory API shenanigans as a hack.
> > 
> > > > Another issue is the complete lack of endian conversions. Does it
> > > > matter? It depends. Will anyone ever use this on a big endian host and
> > > > move the meta data backing file to a little endian host? Probably not.
> > > > So does it really matter? Probably not, but it is cutting corners.
> > > >
> > 
> > After I had replied this, I considered a follow-up, because there are
> > probably QEMU developers that would call me out on this.
> > 
> > This definitely DOES matter to QEMU.
> > 
> > >
> > > Great point on endianness! Naturally, all file backed values are stored in
> > > their native endianness. This way, there is no extra overhead on big endian
> > > hardware architectures. Portability concerns can be easily addressed by
> > > storing metadata endianness as a byte flag in its header. Then, during
> > > initialization, the metadata validation code can detect the possible
> > > discrepancy in endianness and automatically convert the metadata to the
> > > endianness of the host. This part is out of scope of this series, but I would
> > > be able to contribute such a solution as an enhancement in the future.
> > >
> > 
> > It is not out of scope. I don't see why we should merge something that
> > is arguably buggy.
> 
> Again, wow! Now you turned around and arbitrarily elevated this issue from
> moderate ("Does it matter?, cutting corners") to severe ("buggy"). Likely
> because v5 of WDC patchset has been posted.

No, exactly as I wrote above, after I hit reply I considered a
follow-up. I guess I should have.

> This, again, just shows your lack of integrity as a maintainer.
> 

I can't believe I just read that.

I will not put up with this. It is completely non-called for. I stand up
for my opinions and I will fight to make sure the best possible code
goes upstream. Yes, I am paid by Samsung. But I can compartmentalize. I
have been working on QEMU before Samsung and I know how to separate
corporate interest and open source. I have a proven record on this list
to show that. I really cannot believe that you brought it down to that
level. I have been putting forth technical arguments throughout this
entire review process and now you are getting personal.

Not. Cool. Please keep things professional from now.

> This "issue" is a real trivial one to fix as I described above and you are
> blowing it up way out of proportion, making it look like it is a fundamental
> problem that can not be resolved. It's not.
> 

If it is so trival to fix, please fix it. I think I made it clear that I
won't be happy until it is portable.

And please note that I have *not* complained about other parts of your
series. I have complained ALOT about the persistence implementation -
and I continue to stand behind those complaints.

I'm getting super tired of this one-sided process. I have continuously
reviewed and commented your series, I have found multiple bugs, I have
suggested improvements. Maybe if just one or two of those 9 people who
signed off on your zoned implementation could look past their own nose
and look at my series - you might just realize that its decent, portable
and offers some niceties that yours do not have (at the cost of the same
amount of code mind you).

All,

Let's de-escalate this, please. There's no reason to doubt Klaus wants
to see this to work well, just as everyone else does. We unfortunately
have conflicting proposals posted, and everyone is passionate enough
about their work, but please simmer down.

As I mentioned earlier, I'd like to refocus on the basic implementation
and save the persistent state discussion once the core is solid. After
going through it all, I feel there's enough to discuss there to keep us
busy for little while longer. Additional comments on the code will be
coming from me later today.

> -----Original Message-----
> From: Keith Busch <kbusch@kernel.org>
> Sent: Tuesday, September 29, 2020 3:22 PM
> To: Klaus Jensen <its@irrelevant.dk>
> Cc: Dmitry Fomichev <Dmitry.Fomichev@wdc.com>; Kevin Wolf
> <kwolf@redhat.com>; Fam Zheng <fam@euphon.net>; Damien Le Moal
> <Damien.LeMoal@wdc.com>; qemu-block@nongnu.org; Niklas Cassel
> <Niklas.Cassel@wdc.com>; Klaus Jensen <k.jensen@samsung.com>; qemu-
> devel@nongnu.org; Alistair Francis <Alistair.Francis@wdc.com>; Philippe
> Mathieu-Daudé <philmd@redhat.com>; Matias Bjorling
> <Matias.Bjorling@wdc.com>
> Subject: Re: [PATCH v4 00/14] hw/block/nvme: Support Namespace Types
> and Zoned Namespace Command Set
> 
> All,
> 
> Let's de-escalate this, please. There's no reason to doubt Klaus wants
> to see this to work well, just as everyone else does. We unfortunately
> have conflicting proposals posted, and everyone is passionate enough
> about their work, but please simmer down.
> 
> As I mentioned earlier, I'd like to refocus on the basic implementation
> and save the persistent state discussion once the core is solid. After
> going through it all, I feel there's enough to discuss there to keep us
> busy for little while longer. Additional comments on the code will be
> coming from me later today.

OK, I agree with this and I will not be replying to the email prior to this
one it the thread. Let's calm down so we will be able to have a beer at a
conference one day :)

The only one thing that I would like to cover is lack of response to Klaus'
ZNS patchset. Klaus, you are right to complain about it. Since discovering
about the large backlog of NVMe patches that you had pending
(something that we were not aware at the time of publishing our patches),
we made the decision to rebase our series on top of the patches that you
had posted before the publication time of WDC ZNS patchset. Since then,
I got caught in the constant cycle of rebasing our patches on top of your
series and that prevented me from doing much in terms of reviewing of
your commits. Now, once we seem to catch up with the current head of
development, I should be able to do more of this. There is absolutely no
ill will involved :)

Dmitry