... into snd_emu1010_load_firmware_entry(). This makes it clearer that
these steps belong together tightly, as implied by prior commits.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Message-ID: <20240428093717.3198716-5-oswald.buddenhagen@gmx.de>
It is a low-level I/O access function, so io.c is the natural place for
it.
While we're moving the code, reduce the scope of some variables, use
compound assignment operators, and add/adjust some comments.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Message-ID: <20240428093717.3198716-4-oswald.buddenhagen@gmx.de>
The FPGA access through the GPIO port does not interfere with other
sound processor register access, so there is no need to subject it to
emu_lock. And after moving all FPGA access out of the interrupt handler,
it does not need to be IRQ-safe, either.
What's more, attaching the dock causes a firmware upload, which takes
several seconds. We really don't want to disable IRQs for this long, and
even less also have someone else spin with IRQs disabled waiting for us.
Therefore, use a mutex for FPGA access locking.
This makes the code somewhat more noisy, as we need to wrap bigger
sections into the mutex, as it needs to enclose the spinlocks.
The latter has the "side effect" of fixing dock FPGA programming in a
corner case: a really badly timed mixer access right between entering
FPGA programming mode and uploading the netlist would mess up the
protocol.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Message-ID: <20240428093716.3198666-5-oswald.buddenhagen@gmx.de>
The actual event processing was already done by workqueue items. We can
move the event dispatching there as well, rather than doing it already
in the interrupt handler callback.
This change has a rather profound "side effect" on the reliability of
the FPGA programming: once we enter programming mode, we must not issue
any snd_emu1010_fpga_{read,write}() calls until we're done, as these
would badly mess up the programming protocol. But exactly that would
happen when trying to program the dock, as that triggers GPIO interrupts
as a side effect. This is mitigated by deferring the actual interrupt
handling, as workqueue items are not re-entrant.
To avoid scheduling the dispatcher on non-events, we now explicitly
ignore GPIO IRQs triggered by "uninteresting" pins, which happens a lot
as a side effect of calling snd_emu1010_fpga_{read,write}().
Fixes: fbb64eedf5 ("ALSA: emu10k1: make E-MU dock monitoring interrupt-driven")
Link: https://bugzilla.kernel.org/show_bug.cgi?id=218584
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Message-ID: <20240428093716.3198666-4-oswald.buddenhagen@gmx.de>
Resulting from more reverse engineering in the course of debugging.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Message-ID: <20240406064830.1029573-15-oswald.buddenhagen@gmx.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The hardware supports S16LE and U8 samples, while U16LE and S8 (which
the driver implicitly claims to support) require sign flipping.
Note that this matters only for the GUS patch loader, as the implemented
SoundFont v2.01 spec is limited to S16LE.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Message-ID: <20240406064830.1029573-10-oswald.buddenhagen@gmx.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Given that the filter is already set to neutral for PCM voices, the
only observable effect is that the Z1/Z2/FXBUS registers don't have a
stray bit set for negative numbers anymore. The bit is below the ones
significant for output, but it would mess with 32-bit sample
recombination, which we intend to add.
kX-project does that, but I had to figure out myself why.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230715160802.326872-1-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
85;95;0c
This uses IRQs to track spontaneous changes to the word clock source
register.
FWIW, that this can happen in the first place is the reason why it is
futile to lock the clock source mixer setting while the device is open -
we can't consistently control the rate anyway. Though arguably, we
should reset any open streams when that happens, as they become
corrupted anyway.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230715160738.326832-1-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Now that we know the actual word clock, we can:
- Put the resulting rate into the hardware info
- At 44.1 kHz word clock shift the rate for the pitch calculations,
which presume a 48 kHz word clock
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230612191325.1315854-5-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The value isn't used yet; the subsequent commits will do that.
This ignores the existence of rates above 48 kHz, which is fine, as the
hardware will just switch to the fallback clock source when fed with a
rate which is incompatible with the base clock multiplier, which
currently is always x1.
The sample rate display in /proc spdif-in is adjusted to reflect our
understanding of the input rates.
This is tested only with an 0404b card without sync card, so there is a
lot of room for improvement.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230612191325.1315854-4-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The actually available clock sources depend on the available audio input
ports and dedicated clock input ports.
This includes refactoring the code to be data-driven to remain
manageable.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230612191325.1315854-3-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
So far, we set the fallback as a side effect of setting the source. But
the fallback makes no sense at all when an internal clock is selected.
Defaulting to 48k for S/PDIF & ADAT makes sense, but as that is the
global default and we're not changing it automatically any more, it's
just fine to leave it entirely to the explicit setting.
This changes the name of the pre-existing control to something more
appropriate (regardless of the split), so users will need to adjust
their mixer settings.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230612191325.1315854-2-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
- Include the FX bus map, without which the already present send routing
info would require looking up the documentation.
- Include the physical I/O channels as known to the driver
- Make the multi-channel capture map actually name the mapped input
channels rather than "FXBUS" (Audigy) or even "???" (SbLive)
- The latter two are omitted for E-MU cards, as their physical I/O is
routed through the FPGA
- While at it, make the "Card" field somewhat more useful
This includes de-duplicating the label tables between emuproc and emufx,
updating/improving the FX bus label table, and making the SB Live! 5.1
multi-track capture channel mapping hack data-driven.
Tested-by: Jonathan Dowland <jon@dow.land>
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230526101659.437969-7-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Include the routing information, which can be actually read back.
Somewhat as a drive-by, make the register dump format less obscure - the
previous one made no sense at all.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230526101659.437969-6-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
We use independent voices for the channels, so we need to make an effort
to ensure that they are actually in sync.
The hardware doesn't provide atomicity, so we may need to retry a few
times, due to NMIs, PCI contention, and the wrong phase of the moon.
Solution inspired by kX-project.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230523200709.236023-3-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Instead of separate voices, we now allocate non-interleaved channels,
which may in turn contain two interleaved voices each. The higher-level
code keeps only one pointer per channel. The channels are not allocated
in one block any more, as there is no reason to do that. As a
consequence of that, and because it is cleaner regardless, we now let
the allocator store these pointers at a specified location, rather than
returning only the first one and having the calling code deduce the
remaining ones.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230518140947.3725394-8-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The voice allocator clearly knows about the field (it resets it), so
it's more consistent (and leads to less duplicated code) to have the
constructor take it as a parameter.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230518140947.3725394-7-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This allows us to drop the code that tries to preserve already allocated
voices upon repeated hw_param callback invocations. Getting it right for
multi-channel voices would otherwise get a bit hairy.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230518140947.3725394-5-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Eliminate the MIDI type, as there is no such thing - the MPU401 port
doesn't have anything to do with voices.
For clarity, differentiate between regular and extra voices.
Don't atomize the enum into bits in the table display.
Simplify/optimize the storage.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230518140947.3725394-4-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
On Audigy, the send amounts are merely targets, presumably to avoid
sound distortion due to sudden changes, which the EMU8K docu explicitly
warns about.
However, that "soft-start" would prevent bit-for-bit reproduction, so
we now force the current send amounts to their final values at PCM
playback init.
One might want to do that for the MIDI synthesizer as well, though it
seems mostly pointless due to the attack phase each note has anyway.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230518140339.3722279-3-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
While this nicely denoises the code, the real intent is being able to
write many registers pseudo-atomically, which will come in handy later.
Idea stolen from kX-project.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230518093134.3697955-1-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The cache causes a fixed delay regardless of stream parameters.
Consequently, all that "cache invalidate size" calculation stuff was
garbage (which can be traced right back to Creative's OSS driver).
This also removes the definitions of registers CD1..CDF, because they
are accessed only relative to CD0 anyway.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230517174256.3657060-5-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This workaround fails to address the underlying problem, which is
actually wholly self-made. Subsequent patches will fix it.
This reverts commit 56385a12d9.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230517174256.3657060-2-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Shrink the {in,out}put_source arrays and their data type to what is
actually necessary.
To be still on the safe side, add some static asserts.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230516093612.3536508-11-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Unlike the other models, this is actually a distinct card, rather than
an E-MU 1010 with different "dongles". It is stereo only, and supports
no ADAT (there is no trace of ADAT in the manual, switching the output
mode to ADAT has no effect, and switching the input mode to ADAT just
breaks input (presumably ... my only ADAT source is the card's output)).
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230516093612.3536508-10-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Define arrays of strings instead of snd_kcontrol_new.
While at it, move the E-MU source & destination enum defs next to their
hardware defs, which is a lot more logical and will come in handy in a
followup commit. And add some static asserts to verify that the array
sizes match.
This also applies the compactization from the previous commit to the
destination registers.
While reshuffling the arrays anyway, switch the order of the HAMOA_DAC
& HANA_SPDIF output destinations for the 1010 card, so they follow a
more regular pattern. This should have no functional impact.
The code is somewhat de-duplicated by the extraction of add_ctls().
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230516093612.3536508-4-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Many registers are meaningless for stereo slaves and the extra voices.
This patch cleans up these unnecessary register writes.
snd_emu10k1_playback_{trigger,stop}_voice() is not called for stereo
slaves any more.
snd_emu10k1_playback_prepare_voice() is renamed to
snd_emu10k1_playback_unmute_voice(), as this better reflects its
remaining function. It's not called for the extra voices any more.
Accordingly, snd_emu10k1_playback_mute_voice() is factored out from
snd_emu10k1_playback_stop_voice(), and is called selectively as well.
This doesn't add conditionals which would avoid initializing
sub-registers, as that wouldn't pull its weight.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230516093612.3536451-6-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
We initialize them at card init and don't touch them later, so there is
no need to reset them again at voice start.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230516093612.3536451-4-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
We (rightfully) don't enable the envelope engine for PCM voices, so any
related setup is entirely pointless - the EMU8K documentation makes that
very clear, and the fact that the various open drivers all use different
values to no observable detriment pretty much confirms it.
The remaining initializations are regrouped for clarity.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230516093612.3536451-3-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The voice volume is a raw fractional multiplier that can't actually
represent 1.0. To still enable real pass-through, we now set the volume
to 0.5 (which results in no loss of precision, as the FX bus provides
fractional values) and scale up the samples in DSP code.
To maintain backwards compatibility with existing configuration files,
we rescale the values in the mixer controls. The range is extended
upwards from 0xffff to 0x1fffd, which actually introduces the
possibility of specifying an amplification.
There is still a minor incompatibility with user space, namely if
someone loaded custom DSP code. They'll just get half the volume, so
this doesn't seem like a big deal.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230514170323.3408834-8-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Fractional multiplication with the maximal value 2^31-1 causes some tiny
distortion. Instead, we want to multiply with the full 2^31. The catch
is of course that this cannot be represented in the DSP's signed 32 bit
registers.
One way to deal with this is to encode 1.0 as a negative number and
special-case it. As a matter of fact, the SbLive! code path already
contained such code, though the controls never actually exercised it.
A more efficient approach is to use negative values, which actually
extend to -2^31. Accordingly, for all the volume adjustments we now use
the MAC1 instruction which negates the X operand.
The range of the controls in highres mode is extended downwards, so -1
is the new zero/mute. At maximal excursion, real zero is not mute any
more, but I don't think anyone will notice this behavior change. ;-)
That also required making the min/max/values in the control structs
signed. This technically changes the user space interface, but it seems
implausible that someone would notice - the numbers were actually
treated as if they were signed anyway (and in the actual mixer iface
they _are_). And without this change, the min value didn't even make
sense in the first place (and no-one noticed, because it was always 0).
Tested-by: Jonathan Dowland <jon@dow.land>
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230514170323.3408834-7-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The idea to encode the bitfield manipulation in the register address is
quite clever, but doing that by hand is ugly and error-prone. So derive
it automatically from the mask instead.
Macros cannot #define other macros, so we now declare enums instead.
This also adds macros for decoding the register definitions. These will
be used by later commits.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230514170323.3408798-1-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Evidently, the channel delay bug exists in all E-MU cards; it's in the
Hana FPGA program, and was never fixed.
Note that the implementation is somewhat lazy - to localize the code
paths, we actually waste a GPR and a DSP instruction by keeping two
delay registers for the same physical source.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230510173917.3073107-6-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Firstly, fix the distribution between public and private headers.
Otherwise, some of the already public macros wouldn't actually work, and
the SNDRV_EMU10K1_IOCTL_DBG_READ result for Audigy would be useless.
Secondly, add condition code registers for Audigy. These are just
aliases for selected constant registers, and thus are generation-
specific. At least A_CC_REG_ZERO is actually correct ...
Finally, shuffle around some defines to more logical places while at it,
and fix up some more comments.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230422161021.1143903-7-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
For documentation purposes and later use.
Some pre-existing but (mostly) unused definitions were renamed for
consistency.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230422161021.1143903-6-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Firstly, remove the FXWC_* defines - the comment on FXWC implies that
the relevant defines are the (A_)EXTOUT_* ones. It's unclear where this
came from - it was in the initial ALSA import, but neither the driver
from Creative nor kX-project have these defines.
Secondly, remove A_HR, which made plain no sense (was unused, and
clashed with FXRT). Amends commit cbb7d8f9b7 ("emu10k1: Update
registers defines for the Audigy 2/emu10k2.5").
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230422161021.1143903-5-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Move comments to better locations, de-duplicate, fix/remove incorrect/
outdated ones, add new ones, and unify spacing somewhat.
While at it, also add testing credits for Jonathan Dowland (SB Live!
Platinum) and myself (E-MU 0404b).
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230422161021.1143903-2-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Unlike the Alice2 chips used on 1st generation E-MU cards, the
Tina/Tina2 chips used on the 2nd gen cards have only six GPIN pins,
which means that we need to use a smaller mask. Failure to do so would
falsify the read data if the FPGA tried to raise an IRQ right at that
moment. This wasn't a problem so far, as we didn't actually enable FPGA
IRQs, but that's going to change soon.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230422132430.1057490-1-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
As the register definition clearly states, this is a 16-bit register,
yet we did all accesses as 32-bit. The writes in particular would have
the potential to clear the TIMER register (depending on how the bus/card
actually handles the too long writes).
This commit also introduces a separate define A_GPIO which aliases
A_IOCFG, which better reflects the distinct usage on E-MU cards.
This is done in the same commit to keep the churn down, as we're
touching all involved lines anyway.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230421141006.1005539-2-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Assert the validity of the registers and values, as them being out of
range would indicate an error in the driver. Consequently, don't bother
returning error codes; they were ignored everywhere anyway.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230421141006.1005539-1-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Detach it better from the main PCM driver, which it really doesn't have
much in common with.
In particular, this moves the interrupt handler implementation into
p16v.c, and makes it access the substream runtime status more directly,
so it doesn't need to abuse structs snd_emu10k1_pcm and
snd_emu10k1_voice any more.
We don't need private pcm runtime data at all, as the only thing it was
used for (except the back-link to the substream) was the `running` flag.
So store that directly in runtime->private_data.
This somewhat radical strip-down shows that this driver contains some
complexity that was never actually utilized. I suppose the right way to
fully utilize the hardware in a simple way would be introducing more
substreams. This wouldn't require any of the removed code.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Link: https://lore.kernel.org/r/20230421141006.1005452-7-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
