[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Embedded-pv-devel] [PATCH v15] This is the ABI for the two halves of a para-virtualized sound driver to communicate with each to other.

To: Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>
From: Oleksandr Andrushchenko <andr2000@xxxxxxxxx>
Date: Fri, 27 Jan 2017 17:50:36 +0200
Cc: lars.kurth@xxxxxxxxxx, iurii.konovalenko@xxxxxxxxxxxxxxx, vlad.babchuk@xxxxxxxxx, ian.jackson@xxxxxxxxxxxxx, dario.faggioli@xxxxxxxxxx, tim@xxxxxxx, julien.grall@xxxxxxx, andrii.anisov@xxxxxxxxx, olekstysh@xxxxxxxxx, embedded-pv-devel@xxxxxxxxxxxxxxxxxxxx, al1img@xxxxxxxxx, david.vrabel@xxxxxxxxxx, JBeulich@xxxxxxxx, xen-devel@xxxxxxxxxxxxxxxxxxxx, oleksandr.dmytryshyn@xxxxxxxxxxxxxxx, joculator@xxxxxxxxx
Delivery-date: Fri, 27 Jan 2017 15:50:44 +0000
List-id: <embedded-pv-devel.lists.xenproject.org>

thank you for comments, please find answers below

Can we please switch to v16 discussion as v15 vs v16 is
a big change?

On 01/27/2017 05:14 PM, Konrad Rzeszutek Wilk wrote:

On Thu, Jan 26, 2017 at 12:02:49PM +0200, Oleksandr Andrushchenko wrote:

Hi, Konrad!

First of all thank you very much for the valuable comments

and your time!

The number of changes (mostly in description) is going to

be huge, so do you think I can publish something like

"RFC v16" so we can discuss the updated patch?

RFC sadly means folks are going to mostly ignore it.
I would prefer you did not use RFC at this stage but just
did v16.
..snip..

sure

+ * Example for the frontend running in domain 5, instance of the driver
+ * in the front is 0 (single or first PV driver), device id 2,
+ * first stream (0):
+ * /local/domain/<frontend_id>/device/vsnd/<drv_idx>/
+ *         device/<dev_id>/stream/<stream_idx>/type = "p"
+ * /local/domain/5/device/vsnd/0/device/2/stream/0/type = "p"

Why do you need 'device' ?

just for clarity: the hierarchy of the sound driver would
be that a device may have multiple different streams.

And it just occured to me that you could also imply that
each device has an stream without the 'stream' in it.

So

/local/domain/5/device/vsnd/0/2/0/type = "p"

And the format is:
/local/domain/<front-id>/device/vsnd/<instance of PV 
driver>/<device-id>/<stream-id>

ok, so we'll end up with something like:

--------------------------------- Backend-----------------------------------


/local/domain/0/backend/vsnd/1/0/frontend-id = "1"
/local/domain/0/backend/vsnd/1/0/frontend = "/local/domain/1/device/vsnd/0"
/local/domain/0/backend/vsnd/1/0/state = "4"
/local/domain/0/backend/vsnd/1/0/versions = "1,2"

----------------------------- Card ----------------------------

/local/domain/1/device/vsnd/0/version = "1"
/local/domain/1/device/vsnd/0/short-name = "Card short name"
/local/domain/1/device/vsnd/0/long-name = "Card long name"
/local/domain/1/device/vsnd/0/sample-rates = "8000,32000,44100,48000,96000"
/local/domain/1/device/vsnd/0/sample-formats = "s8,u8,s16_le,s16_be"
/local/domain/1/device/vsnd/0/buffer-size = "262144"

----------------------------- PCM device 0 ----------------------------

/local/domain/1/device/vsnd/0/0/name = "General analog"
/local/domain/1/device/vsnd/0/0/channels-max = "5"

----------------------------- Stream 0, playback----------------------------


/local/domain/1/device/vsnd/0/0/0/type = "p"
/local/domain/1/device/vsnd/0/0/0/sample-formats = "s8,u8"
/local/domain/1/device/vsnd/0/0/0/unique-id = "0"
/local/domain/1/device/vsnd/0/0/0/ring-ref = "386"
/local/domain/1/device/vsnd/0/0/0/event-channel = "15"

------------------------------- PCM device 3--------------------------------


/local/domain/1/device/vsnd/0/3/name = "HDMI-0"
/local/domain/1/device/vsnd/0/3/sample-rates = "8000,32000,44100"

------------------------------ Stream 0, capture----------------------------


/local/domain/1/device/vsnd/0/3/0/type = "c"
/local/domain/1/device/vsnd/0/3/0/unique-id = "2"
/local/domain/1/device/vsnd/0/3/0/ring-ref = "387"
/local/domain/1/device/vsnd/0/3/0/event-channel = "151"

Is this what you would like to see?
IMO, all these values do not help understanding what it is, e.g.
this is equal to me if we have

/local/domain/1/device/vbd/51712/queue-0/ring-ref = "8"
/local/domain/1/device/vbd/51712/queue-0/event-channel = "3"
/local/domain/1/device/vbd/51712/queue-1 = ""
/local/domain/1/device/vbd/51712/queue-1/ring-ref = "9"

and then decided to go with

/local/domain/1/device/vbd/51712/0/ring-ref = "8"
/local/domain/1/device/vbd/51712/0/event-channel = "3"
/local/domain/1/device/vbd/51712/1/ring-ref = "9"

Can one easily tell what 0 or 1 after "51712/" is?

So, what is the final decision then?

Though I have a little of trouble with the 'instance of the
driver'. Are you suggesting you would have multiple
PV drivers of 'vsnd'? Can't the multiple device ids fulfill this?

it is possible, but the main use-case will have a single
PV driver with multiple PCM devices/streams

So, from readability POV I would still have "device" in place
 From xenstore documentation: "Data should generally be
human-readable for ease of management and debugging "
I assume this also applies to the structure as well

Could not this be:

/local/domain/5/device/vsnd/0/2/stream/0/type = "p" ?

then one has to know that "2" stands for device.
see above, I would keep "device" here

+ *
+ *------------------------------- PCM settings --------------------------------
+ *
+ * Every virtualized sound frontend has set of devices and streams, each

frontend or backend?

I would think backend since this is still the backend section?

you are right, moved to frontend's section

+ * is individually configured. Part of the PCM configuration can be defined at
+ * higher level and be fully or partially re-used by the underlying layers.
+ * These configuration values are:
+ *  o number of channels (min/max)
+ *  o supported sample rates
+ *  o supported sample formats.
+ * E.g. one can define these values for the whole driver, device or stream.
+ * Every underlying layer in turn can re-define some or all of them to better
+ * fit its needs. For example, driver may define number of channels to be
+ * in [1; 8] range, and some particular stream may be limited to [1; 2] only.
+ * The rule is that the underlying layer must be a subset of the upper layer
+ * range.

OK, so these are more of 'global ones'?

kind of, what is more they can be "global" to all the
underlying levels, e.g. if device changes some of the
parameters they become adopted by streams if not
explicitly configured. The same is true if device changes
parameters set at the card level.
This way one can easily tailor particular pieces of the
system w/o writing all settings again

Are they "global" to the "instance of the PV drivers" or all the
device ids?

they are global to the frontend and the settings
live on frontend's xenstore path. Please check
configuration example in v16


I presume the latter which is why you need to have multiple
instances of the PV drivers support?

(You may want to mention this requirement in the section talking
about the format of how the structure is exposed in the backend
and frontend).

I've already put full configuration xenstore tree with
values for education purposes, so one can see how it gets
configured

+ *
+ * Note: if any of the values are not defined then PV driver should use
+ * its default values instead.

How is that suppose to work? What if the PV driver by default wants
to use say 314156 channels and the backend does not expose any? Can
it use that? I presume it could try.. and it would fail but that
hardly seems nice.

I will probably remove this sentence and the below at all.
If not defined this could mean an error state

+ *
+ * channels-min
+ *      Values:         <uint>

uint32_t

+ *
+ *      The minimum amount of channels that is supported.
+ *      Must be at least 1. If not defined then use frontend's default.

This is odd. This whole section is for the backend, so how can the backend
not have this defined?

Do you mean to say that this parameter is optional? Perhaps
then say right before 'The minimum' include '(optional)'

Anyhow where do you envision this optional global value to be?

/local/domain/0/backend/vsnd/5/channels-[min|max] ?

Perhaps mention that in the start of this section:

These global values are under the backend 'vsnd' directory, as
so:

/local/domain/0/backends/vsnd/<front-id>/channels-[min|max]

Thought I think there is probably a better way to
say this..

well, the confusion comes from the fact that I have
described all this under backend's section
now when I move it to the frontend's section it should
make more sense

OK, but do please have an 'backend' section.

done,
I have clearly(?) defined both frontend and backend entries in v16

+ *
+ * channels-max
+ *      Values:         <uint>

uint32_t

uint8_t

+ *
+ *      The maximum amount of channels that is supported.
+ *      Must be at least <channels-min>. If not defined then use frontend's
+ *      default.

I am still having trouble understanding how the backend is suppose
to use the frontend's default.

Perhaps you want to say:

If this value is not exposed by the backend the frontend
is permitted to use its default values.

?

moved to frontend's section

+ *
+ * sample-rates
+ *      Values:         <list of uints>

.. of uint32_t

done

+ *
+ *      List of supported sample rates separated by XENSND_LIST_SEPARATOR.
+ *      If not defined then use frontend's default. Sample rates are expressed

Ahain this 'If not defined then use fronten'ds default' could be written
a bit differently.

moved to frontend's section

+ *      as a list of decimal values w/o any ordering requirement.
+ *
+ * sample-formats
+ *      Values:         <list of XENSND_PCM_FORMAT_XXX_STR>
+ *
+ *      List of supported sample formats separated by XENSND_LIST_SEPARATOR.
+ *      If not defined then use frontend's default.

Again, this does not sound right. The backend has no clue what
the frontend defaults are.

moved to frontend's section

+ *
+ * buffer-size
+ *      Values:         <uint>

uint32_t

done

+ *
+ *      The maximum size in octets of the buffer to allocate per stream.
+ *
+ * Example configuration:
+ *
+ * Driver configuration used by all streams:
+ * /local/domain/5/device/vsnd/0/sample-formats = "s8;u8;s16_le;s16_be"

Huh? But this is a backend value? Don't you mean:

/local/domain/0/backend/vsnd/5/sample-formats = "s8;u8;s16_le;s16_be"
?

moved to frontend's section

+ * Stream overrides sample rates supported:

-EPARSE

done

+ * /local/domain/5/device/vsnd/0/device/2/stream/0/sample-rates =
+ *        "8000;22050;44100;48000"

/local/domain/0/backend/vsnd/5/2/stream/0/sample-rates ="8000;22050;44100;48000"

But that is wrong as this section is about global values so they
would be more like:

/local/domain/0/backend/vsnd/5/sample-rates ="8000;22050;44100;48000"

?

not anymore, as I move these to frontend's configuration

+ *
+ *----------------------- Virtual sound card settings 
--------------------------
+ * short-name
+ *      Values:         <char[32]>
+ *
+ *      Short name of the virtual sound card. Optional.
+ *
+ * long-name
+ *      Values:         <char[80]>
+ *
+ *      Long name of the virtual sound card. Optional.
+ *
+ * For example,
+ * /local/domain/5/device/vsnd/0/short-name = "Virtual audio"
+ * /local/domain/5/device/vsnd/0/long-name =
+ *         "Virtual audio at center stack"

This should be inverted. That is this is on the backend
side:

/local/domain/0/backend/vsnd/5/0/2/short-name = "Virtual audio"

And so on.

Also how come your example is missing the device part?

I will put a complete configuration example

Wait, why do you have an an device below but ... not here?
Why not group them together?

That is why not have 'short-name' ,'long-name' and 'name' in
the same directory?

name is for a stream; short/long for the card

+ *
+ *----------------------------- Device settings 
--------------------------------
+ * name
+ *      Values:         <char[80]>
+ *
+ *      Name of the sound device within the virtual sound card. Optional.
+ *
+ * For example,
+ * /local/domain/5/device/vsnd/0/device/0/name = "General analog"

The example you used at start had a device value of 2. It may
be better to use that here, so :

/local/domain/0/backend/vsnd/5/2/name = "General analog"

done

+ *
+ *----------------------------- Stream settings -------------------------------
+ *
+ * type
+ *      Values:         "p", "c"
+ *
+ *      Stream type: "p" - playback stream, "c" - capture stream
+ *
+ *      If both capture and playback are needed then two streams need to be
+ *      defined under the same device. For example,
+ *      /local/domain/5/device/vsnd/0/device/0/stream/0/type = "p"
+ *      /local/domain/5/device/vsnd/0/device/0/stream/1/type = "c"

OK, but again:
/local/domain/0/backend/vsnd/5/2/0/stream/0/type = "p"
/local/domain/0/backend/vsnd/5/2/0/stream/1/type = "c"

Do you want also an 'feature-version' to identify what version
the backend exposes?

will add that

+ *
+ *****************************************************************************
+ *                            Frontend XenBus Nodes
+ *****************************************************************************
+ *
+ *----------------------- Request Transport Parameters -----------------------
+ *
+ * These are per stream.

.. which means they are also per device right?

So one can have something like this:

/local/domain/5/device/vsnd/2/stream/0/event-channel
/local/domain/5/device/vsnd/0/stream/0/event-channel
/local/domain/5/device/vsnd/2/stream/1/event-channel
/local/domain/5/device/vsnd/0/stream/0/event-channel

yes, every stream has its own communication channel

+ *
+ * event-channel
+ *      Values:         <uint>

uint32_t

done

+ *
+ *      The identifier of the Xen event channel used to signal activity
+ *      in the ring buffer.
+ *
+ * ring-ref
+ *      Values:         <uint>

uint32_t

done

+ *
+ *      The Xen grant reference granting permission for the backend to map
+ *      a sole page in a single page sized ring buffer.
+ *
+ * index

Why not 'unique-id' ?

makes sense, will change

+ *      Values:         <uint>

uint32_t

done

+ *
+ *      After stream initialization it is assigned a unique ID (within the 
front
+ *      driver), so every stream of the frontend can be identified by the
+ *      backend by this ID. This is not equal to stream_idx as the later is

s/stream_idx/steam-idx/

done

+ *      zero based within a device, but this index is contiguous within the

contingous

done

+ *      driver.
+ */
+
+/*
+ * STATE DIAGRAMS
+ *
+ *****************************************************************************
+ *                                   Startup                                 *
+ *****************************************************************************
+ *
+ * Tool stack creates front and back state nodes with initial state
+ * XenbusStateInitialising.
+ * Tool stack creates and sets up frontend sound configuration nodes per 
domain.
+ *
+ * Front                                Back
+ * =================================    =====================================
+ * XenbusStateInitialising              XenbusStateInitialising
+ *                                       o Query backend device identification
+ *                                         data.
+ *                                       o Open and validate backend device.
+ *                                                      |
+ *                                                      |
+ *                                                      V
+ *                                      XenbusStateInitWait
+ *
+ * o Query frontend configuration
+ * o Allocate and initialize
+ *   event channels per configured
+ *   playback/capture stream.
+ * o Publish transport parameters
+ *   that will be in effect during
+ *   this connection.
+ *              |
+ *              |
+ *              V
+ * XenbusStateInitialised
+ *
+ *                                       o Query frontend transport parameters.
+ *                                       o Connect to the event channels.
+ *                                                      |
+ *                                                      |
+ *                                                      V
+ *                                      XenbusStateConnected
+ *
+ *  o Create and initialize OS
+ *  virtual sound device instances
+ *  as per configuration.
+ *              |
+ *              |
+ *              V
+ * XenbusStateConnected
+ *
+ *                                      XenbusStateUnknown
+ *                                      XenbusStateClosed
+ *                                      XenbusStateClosing
+ * o Remove virtual sound device
+ * o Remove event channels
+ *              |
+ *              |
+ *              V
+ * XenbusStateClosed
+ *
+ */
+
+/*
+ * PCM FORMATS
+ *
+ * XENSND_PCM_FORMAT_<format>[_<endian>]
+ *
+ * format: <S/U/F><bits> or <name>
+ *     S - signed, U - unsigned, F - float
+ *     bits - 8, 16, 24, 32
+ *     name - MU_LAW, GSM, etc.
+ *
+ * endian: <LE/BE>, may be absent
+ *     LE - Little endian, BE - Big endian
+ */
+#define XENSND_PCM_FORMAT_S8            0
+#define XENSND_PCM_FORMAT_U8            1
+#define XENSND_PCM_FORMAT_S16_LE        2
+#define XENSND_PCM_FORMAT_S16_BE        3
+#define XENSND_PCM_FORMAT_U16_LE        4
+#define XENSND_PCM_FORMAT_U16_BE        5
+#define XENSND_PCM_FORMAT_S24_LE        6
+#define XENSND_PCM_FORMAT_S24_BE        7
+#define XENSND_PCM_FORMAT_U24_LE        8
+#define XENSND_PCM_FORMAT_U24_BE        9
+#define XENSND_PCM_FORMAT_S32_LE        10
+#define XENSND_PCM_FORMAT_S32_BE        11
+#define XENSND_PCM_FORMAT_U32_LE        12
+#define XENSND_PCM_FORMAT_U32_BE        13
+#define XENSND_PCM_FORMAT_F32_LE        14 /* 4-byte float, IEEE-754 32-bit, */
+#define XENSND_PCM_FORMAT_F32_BE        15 /* range -1.0 to 1.0              */
+#define XENSND_PCM_FORMAT_F64_LE        16 /* 8-byte float, IEEE-754 64-bit, */
+#define XENSND_PCM_FORMAT_F64_BE        17 /* range -1.0 to 1.0              */
+#define XENSND_PCM_FORMAT_IEC958_SUBFRAME_LE 18
+#define XENSND_PCM_FORMAT_IEC958_SUBFRAME_BE 19
+#define XENSND_PCM_FORMAT_MU_LAW        20
+#define XENSND_PCM_FORMAT_A_LAW         21
+#define XENSND_PCM_FORMAT_IMA_ADPCM     22
+#define XENSND_PCM_FORMAT_MPEG          23
+#define XENSND_PCM_FORMAT_GSM           24
+
+/*
+ * REQUEST CODES.
+ */
+#define XENSND_OP_OPEN                  0
+#define XENSND_OP_CLOSE                 1
+#define XENSND_OP_READ                  2
+#define XENSND_OP_WRITE                 3
+#define XENSND_OP_SET_VOLUME            4
+#define XENSND_OP_GET_VOLUME            5
+#define XENSND_OP_MUTE                  6
+#define XENSND_OP_UNMUTE                7
+
+/*
+ * XENSTORE FIELD AND PATH NAME STRINGS, HELPERS.
+ */
+#define XENSND_DRIVER_NAME              "vsnd"
+
+#define XENSND_LIST_SEPARATOR           ";"
+/* Path entries */
+#define XENSND_PATH_DEVICE              "device"
+#define XENSND_PATH_STREAM              "stream"
+/* Field names */
+#define XENSND_FIELD_VCARD_SHORT_NAME   "short-name"
+#define XENSND_FIELD_VCARD_LONG_NAME    "long-name"
+#define XENSND_FIELD_RING_REF           "ring-ref"
+#define XENSND_FIELD_EVT_CHNL           "event-channel"
+#define XENSND_FIELD_DEVICE_NAME        "name"
+#define XENSND_FIELD_TYPE               "type"
+#define XENSND_FIELD_STREAM_INDEX       "index"
+#define XENSND_FIELD_CHANNELS_MIN       "channels-min"
+#define XENSND_FIELD_CHANNELS_MAX       "channels-max"
+#define XENSND_FIELD_SAMPLE_RATES       "sample-rates"
+#define XENSND_FIELD_SAMPLE_FORMATS     "sample-formats"
+#define XENSND_FIELD_BUFFER_SIZE        "buffer-size"
+
+/* Stream type field values. */
+#define XENSND_STREAM_TYPE_PLAYBACK     "p"
+#define XENSND_STREAM_TYPE_CAPTURE      "c"
+/* Sample rate max string length */
+#define XENSND_SAMPLE_RATE_MAX_LEN      6

Hm, but it is uint32_t which can be much much higher than 6.

4294967295

11 characters?

fixed

+/* Sample format field values */
+#define XENSND_SAMPLE_FORMAT_MAX_LEN    24

You sure? You may want to make that clear in 'sample-format'
section that the maximum of a string can be 24 characters.

done

And also explain why 24 characters.

no particular reason, just to fit XENSND_PCM_FORMAT_???_STR
do you want me to change it to something else?

+
+#define XENSND_PCM_FORMAT_S8_STR        "s8"
+#define XENSND_PCM_FORMAT_U8_STR        "u8"
+#define XENSND_PCM_FORMAT_S16_LE_STR    "s16_le"
+#define XENSND_PCM_FORMAT_S16_BE_STR    "s16_be"
+#define XENSND_PCM_FORMAT_U16_LE_STR    "u16_le"
+#define XENSND_PCM_FORMAT_U16_BE_STR    "u16_be"
+#define XENSND_PCM_FORMAT_S24_LE_STR    "s24_le"
+#define XENSND_PCM_FORMAT_S24_BE_STR    "s24_be"
+#define XENSND_PCM_FORMAT_U24_LE_STR    "u24_le"
+#define XENSND_PCM_FORMAT_U24_BE_STR    "u24_be"
+#define XENSND_PCM_FORMAT_S32_LE_STR    "s32_le"
+#define XENSND_PCM_FORMAT_S32_BE_STR    "s32_be"
+#define XENSND_PCM_FORMAT_U32_LE_STR    "u32_le"
+#define XENSND_PCM_FORMAT_U32_BE_STR    "u32_be"
+#define XENSND_PCM_FORMAT_F32_LE_STR    "float_le"
+#define XENSND_PCM_FORMAT_F32_BE_STR    "float_be"
+#define XENSND_PCM_FORMAT_F64_LE_STR    "float64_le"
+#define XENSND_PCM_FORMAT_F64_BE_STR    "float64_be"
+#define XENSND_PCM_FORMAT_IEC958_SUBFRAME_LE_STR "iec958_subframe_le"
+#define XENSND_PCM_FORMAT_IEC958_SUBFRAME_BE_STR "iec958_subframe_be"
+#define XENSND_PCM_FORMAT_MU_LAW_STR    "mu_law"
+#define XENSND_PCM_FORMAT_A_LAW_STR     "a_law"
+#define XENSND_PCM_FORMAT_IMA_ADPCM_STR "ima_adpcm"
+#define XENSND_PCM_FORMAT_MPEG_STR      "mpeg"
+#define XENSND_PCM_FORMAT_GSM_STR       "gsm"
+
+/*
+ * STATUS RETURN CODES.
+ */
+/* Operation not supported. */
+#define XENSND_RSP_NOTSUPP              (-2)
+/* Operation failed for some unspecified reason (e. g. -EIO). */

Is that the -EIO going to be exposed somewhere?

no

   Inside of the
'struct xensnd_resp' ?

no, XENSND_RSP_??? is used there

Would it make sense to have it exposed there? Is that what 'status'
is for?

see above, XENSND_RSP_???

Why not have an 'err' and mandate that it use POSIX return values?

well, if we agree on POSIX values then I am more than ok

I think it is a good move going forward.

For example the pciff assumed it would only need these type of
errors and 'err' would have XEN_PCI_ERR_* errors.
And the MSI-X came in and they messed it up and
'err' has now POSIX return values _and_ XEN_PCI_ERR_* depending
on the type of operation.

good

+#define XENSND_RSP_ERROR                (-1)
+/* Operation completed successfully. */
+#define XENSND_RSP_OKAY                 0
+
+/*
+ * Assumptions:
+ *   o usage of grant reference 0 as invalid grant reference:
+ *     grant reference 0 is valid, but never exposed to a PV driver,
+ *     because of the fact it is already in use/reserved by the PV console.
+ *   o all references in this document to page sizes must be treated
+ *     as pages of size XEN_PAGE_SIZE unless  otherwise noted.

                                                  ^- extra space

fixed

+ *
+ * Description of the protocol between frontend and backend driver.

Is this suppose to have an:
-----------------------------------------
underneath it?

done

+ *
+ * The two halves of a Para-virtual sound driver communicate with
+ * each other using a shared page and an event channel.

shared pages and event channels?

Since it looks like  you can have

/local/domain/<frontend-id>/device/vsnd/<device-id>/stream/<stream-id>/event-channel

And you can have N device-id and M stream-id ?

right, done

+ * Shared page contains a ring with request/response packets.
+ *
+ * All reserved fields in the structures below must be 0.
+ *
+ * All request packets have the same length (32 octets)

Which implies you can have at maximum 64 requests?

[64 bytes for the four RING_IDX along with the 48 of padding, that
means 4032 left, but since we need this to be modulo 2 the best
we can do is 2^6.

you mean we have to pad the structures so they are all
64 bytes long?

No no. Just that you can only fit 64 requests on a page.

The first 64 bytes of the ring are for the producer and consumer
index values. Then after that your structures which are up to 32 bytes
are filled out.

But since the entries need to be module two that means the maximum
of these structures you can fill out is 64.

have 32-octet structures in v16

+ * All request packets have common header:
+ *          0                 1                  2                3        
octet
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                 id                |    operation    |     stream_idx  |

Why the stream_idx ? Your 'ring-ref' is rooted from the '<stream-idx>' so
the frontend and backend already know this.

indeed, we can probably remove this from all the structures

+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                                reserved                               |
+ * +-----------------+-----------------+-----------------+-----------------+
+ *   id - uint16_t, private guest value, echoed in response
+ *   operation - uint8_t, operation code
+ *   stream_idx - uint8_t, index of the stream ("streams_idx" XenStore entry
+ *     of the stream)

Why do you need this duplicate information?

it describes common header and its values

I meant the 'stream_idx'

re-worked in v16

+ *
+ *
+ * Request open - open a PCM stream for playback or capture:
+ *          0                 1                  2                3        
octet
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                 id                | XENSND_OP_OPEN  |     stream_idx  |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                                reserved                               |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                                pcm_rate                               |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |  pcm_format     |  pcm_channels   |             reserved              |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                               buffer_sz                               |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                         gref_directory_start                          |

Perhaps 'gref_list' ?

not sure, it is the start of the page directory, its gref

Directory and list are similar enough. If you want directory keep it
but please remove the 'start'. It is implied.

sounds reasonable

+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                               reserved                                |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/|
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                               reserved                                |
+ * +-----------------+-----------------+-----------------+-----------------+
+ *
+ * pcm_rate - uint32_t, stream data rate, Hz
+ * pcm_format - uint8_t, XENSND_PCM_FORMAT_XXX value
+ * pcm_channels - uint8_t, number of channels of this stream

.. And I presume this should be between 'channels-min' and 'channels-max'?
Is it OK to have to say 255 ? What are the boundary values?

yes, it is ok to be uint8_t, IMO. I will change the
xenstore values description as well

+ * buffer_sz - uint32_t, buffer size to be allocated in octets

Is there an maximum? Or a minimum?

Why have it in bytes?  Why not in the amount of grants you need?
Perhaps call it 'gref_nr' ? That way your computation on how
many grants you need is well simplified.

there was a discussion on that:
https://marc.info/?l=xen-devel&m=148008589320533&w=2

You want to capture that in this document (of why you choose
this way).

one would need to read all the story to get to the point, because
lots of iterations were done and some of the info was in between

+ * gref_directory_start - grant_ref_t, a reference to the first shared page

s/gref_directory_start/gref_list/ ?

not sure

+ *   describing shared buffer references. At least one page exists. If shared
+ *   buffer size exceeds what can be addressed by this single page, then

s/shared buffer size/buffer_sz/ ?

why, I am not using variable/field name here, but explaining
I can put the name in brackets)

+ *   reference to the next page must be supplied (see gref_dir_next_page below)

Now what if grefs_nr (or buffer_sz) is say 1 page (4096).
Does that mean that gref_directory_start still needs to point to page
which only has two entries: 0, <grant ref> ?

Or can it be simplified and this gref_directory_start would be used for
data instead?

I think that is what you saying ("If shared .. " which would imply
that if "if !shared" then something else can be done?) but the start
says: "a reference to the first shared page describind shared buffer
references" which implies you do
need this extra indirect page even if the buffer_sz is say 4096.
?

yes, you got it right. the use-cases we have do use buffers
bigger then 4K, so this is why we never thought of such an overhead

+ */
+
+struct xensnd_open_req {

s/_open_req/_req_open/

it will be yet another flame here...
will keep as is

+    uint32_t pcm_rate; /* in Hz */
+    uint8_t pcm_format;
+    uint8_t pcm_channels;
+    uint16_t reserved;
+    uint32_t buffer_sz;
+    grant_ref_t gref_directory_start;
+};
+
+/*
+ * Shared page for XENSND_OP_OPEN buffer descriptor (gref_directory in the
+ *   request) employs a list of pages, describing all pages of the shared data
+ *   buffer:
+ *          0                 1                  2                3        
octet
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                          gref_dir_next_page                           |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                                gref[0]                                |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/|
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                                gref[i]                                |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/|
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                                gref[N -1]                             |
+ * +-----------------+-----------------+-----------------+-----------------+
+ *
+ * gref_dir_next_page - grant_ref_t, reference to the next page describing
+ *   page directory. Must be 0 if no more pages in the list.

s/no more page/there are no more/

done

+ * gref[i] - grant_ref_t, reference to a shared page of the buffer
+ *   allocated at XENSND_OP_OPEN
+ *
+ * Number of grant_ref_t entries in the whole page directory is not
+ * passed, but instead can be calculated as:
+ *   num_grefs_total = (XENSND_OP_OPEN.buffer_sz + XEN_PAGE_SIZE - 1) /
+ *       XEN_PAGE_SIZE

And what are the expected errors? Woudl it make sense to define those?
Say:

   Returns:
   -ENOBUFS: Cannot map that many buffers.
   -EINVAL: Incorrect values in the requst?

ATM, the XENSND_RSP_??? are returned

+ */
+
+struct xensnd_page_directory {
+    grant_ref_t gref_dir_next_page;
+    grant_ref_t gref[1]; /* Variable length */
+};
+
+/*
+ *  Request close - close an opened pcm stream:
+ *          0                 1                  2                3        
octet
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                 id                | XENSND_OP_CLOSE |     stream_idx  |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                               reserved                                |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/|
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                               reserved                                |
+ * +-----------------+-----------------+-----------------+-----------------+
+ */
+
+/*
+ * Request read/write - used for read (for capture) or write (for playback):

Maybe say also that XENSND_OP_OPEN MUST be called before these operations
are permitted.

not sure we need this: normally you have to open something
before you can use it, e.g. a file

+ *          0                 1                  2                3        
octet
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                 id                |    operation    |     stream_idx  |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                               reserved                                |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                                offset                                 |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                                length                                 |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                               reserved                                |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/|
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                               reserved                                |
+ * +-----------------+-----------------+-----------------+-----------------+
+ *
+ * operation - XENSND_OP_READ for read or XENSND_OP_WRITE for write
+ * offset - uint32_t, read or write data offset within the shared buffer
+ *   passed with XENSND_OP_OPEN request

May also want to say that this offset must be less than buffer_sz.

In octets? May want to say that explicitly.

done

+ * length - uint32_t, read or write data length

In octets?

done

+ */
+
+struct xensnd_rw_req {
+    uint32_t offset;
+    uint32_t len;
+};
+
+/*
+ * Request set/get volume - set/get channels' volume of the stream given:
+ *          0                 1                  2                3        
octet
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                 id                |    operation    |     stream_idx  |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                               reserved                                |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/|
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                               reserved                                |
+ * +-----------------+-----------------+-----------------+-----------------+
+ *
+ * operation - XENSND_OP_SET_VOLUME for volume set
+ *   or XENSND_OP_GET_VOLUME for volume get
+ * Buffer passed with XENSND_OP_OPEN is used to exchange volume
+ * values:

Oh. That means you these operation are in effect 'barrier' ones.

As the buffer must be flushed before hand otherwise you would be
overwriting data stream information.

You should probably mention this semantic need?

I think this is implementation specific and shouldn't
be in a generic protocol

How is that implementation specific? If there is something in the page
from the previous command you are overwritting those values.

ok, all the operations are synchronous for the stream given.
it means that if there is something left in the buffer
it will be overwritten by the next req/resp, so this is expected

Or alternatively the XENSND_OP_SET_VOLUME/XENSND_OP_GET_VOLUME could have
a similar structure to 'struct xensnd_rw_req' so that you have
the offset and len?

a page can hold enough values, IMO

Let me see "640K ought to be enough"?

I was talking about 256 channels 4 bytes per each.


You are making assumptions here based on how the implementation
fills out the data structure. But the purpose of the design
is to detach oneself from the implementation and think of
alternative ways.

To capture your words:
"
So if read/write use that buffer, and the volume and muting

controls use it too, how do I change the volume while listening
without disturbing the read/write?

read/write do not happen continuously, e.g. sound card fills its
internal buffers (our buffer is busy) and then until next re-fill our
buffer is free. that means that there is almost no congestion and
always a good chance to set/get volume w/o problem

Jan

"

Well, that is implementation specific. What if some implementaiton
fills it back to back?

I would like you to add the 'offset' and 'len' so that we don't
dig a hole that we can't easily get out of.

ok, I will add
 * +----------------+----------------+----------------+----------------+
 * | offset                               | 12
 * +----------------+----------------+----------------+----------------+
 * | length                               | 16
 * +----------------+----------------+----------------+----------------+
to
1.Request set/get volume - set/get channels' volume of the stream given
2.Request mute/unmute - mute/unmute stream

By this change you enable a use-case when part of the shared buffer
is used for samples and part for volume/mute, right?

+ *
+ *          0                 1                  2                3        
octet
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                               channel[0]                              |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * +/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/|
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                               channel[i]                              |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * +/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/|
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                channel[XENSND_OP_OPEN.pcm_channels - 1]               |
+ * +-----------------+-----------------+-----------------+-----------------+
+ *
+ * channel[i] - sint32_t, volume of i-th channel

You may want to say that i is uint8_t.

done

+ * Volume is expressed as a signed value in steps of 0.001 dB,
+ * while 0 being 0 dB.
+ */
+
+/*
+ * Request mute/unmute - mute/unmute stream:
+ *          0                 1                  2                3        
octet
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                 id                |    operation    |     stream_idx  |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                               reserved                                |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/|
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                               reserved                                |
+ * +-----------------+-----------------+-----------------+-----------------+
+ *
+ * operation - XENSND_OP_MUTE for mute or XENSND_OP_UNMUTE for unmute
+ * Buffer passed with XENSND_OP_OPEN is used to exchange mute/unmute
+ * values:

Again, same comments as above.

+ *
+ *                                     0                                   
octet
+ * +-----------------------------------------------------------------------+
+ * |                               channel[0]                              |
+ * +-----------------------------------------------------------------------+
+ * +/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/|
+ * +-----------------------------------------------------------------------+
+ * |                               channel[i]                              |
+ * +-----------------------------------------------------------------------+
+ * +/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/|
+ * +-----------------------------------------------------------------------+
+ * |                channel[XENSND_OP_OPEN.pcm_channels - 1]               |
+ * +-----------------+-----------------+-----------------+-----------------+
+ *
+ * channel[i] - uint8_t, non-zero if i-th channel needs to be muted/unmuted
+ */
+
+/*
+ * All response packets have the same length (32 octets)
+ *
+ * Response for all requests:
+ *          0                 1                  2                3        
octet
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                 id                |    operation    |     stream_idx  |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |      status     |                      reserved                       |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                               reserved                                |
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/|
+ * +-----------------+-----------------+-----------------+-----------------+
+ * |                               reserved                                |
+ * +-----------------+-----------------+-----------------+-----------------+
+ *
+ * id - uint16_t, copied from the request
+ * stream_idx - uint8_t, copied from request
+ * operation - uint8_t, XENSND_OP_* - copied from request
+ * status - int8_t, response status (XENSND_RSP_*)

Could this become an 'err' and the POSIX values used for errors?

we can discuss this, I am fine with this approach if the
Community accepts this


You can always have _both_.

I am with -XEN_E and int32_t

+ */
+
+struct xensnd_req {
+    uint16_t id;
+    uint8_t operation;
+    uint8_t stream_idx;
+    uint32_t reserved;
+    union {
+        struct xensnd_open_req open;
+        struct xensnd_rw_req rw;
+        uint8_t reserved[24];
+    } op;
+};
+
+struct xensnd_resp {
+    uint16_t id;
+    uint8_t operation;
+    uint8_t stream_idx;
+    int8_t status;
+    uint8_t reserved[27];
+};
+
+DEFINE_RING_TYPES(xen_sndif, struct xensnd_req, struct xensnd_resp);
+
+#endif /* __XEN_PUBLIC_IO_SNDIF_H__ */
--
2.7.4



_______________________________________________
Embedded-pv-devel mailing list
Embedded-pv-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/cgi-bin/mailman/listinfo/embedded-pv-devel

References:
- Re: [Embedded-pv-devel] [PATCH v15] This is the ABI for the two halves of a para-virtualized sound driver to communicate with each to other.
  - From: Oleksandr Andrushchenko

Prev by Date: Re: [Embedded-pv-devel] [Xen-devel] [PATCH v1] displif: add ABI for para-virtual display
Next by Date: Re: [Embedded-pv-devel] [PATCH v15] This is the ABI for the two halves of a para-virtualized sound driver to communicate with each to other.
Previous by thread: Re: [Embedded-pv-devel] [PATCH v15] This is the ABI for the two halves of a para-virtualized sound driver to communicate with each to other.
Next by thread: Re: [Embedded-pv-devel] [PATCH v15] This is the ABI for the two halves of a para-virtualized sound driver to communicate with each to other.
Index(es):
- Date
- Thread

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.