* [PATCH 0/4] ALSA: seq: ump: Fix Note On/Off and key pressure conversion
@ 2026-11-16 3:03 HyeongJun An
2026-11-16 3:03 ` [PATCH 1/4] ALSA: seq: ump: Don't turn a zero key pressure into a Note Off HyeongJun An
` (4 more replies)
0 siblings, 5 replies; 6+ messages in thread
From: HyeongJun An @ 2026-11-16 3:03 UTC (permalink / raw)
To: Takashi Iwai
Cc: Jaroslav Kysela, Mark Brown, Shuah Khan, linux-sound,
linux-kselftest, linux-kernel, HyeongJun An
Hello Takashi,
While extending seq-ump-test, I found three conversion bugs in
seq_ump_convert.c. In each case the UMP core or the other sequencer
path already does the right thing, and the fix follows it.
Patches 1-3 are the fixes. They also apply to for-linus. Patch 4 adds
the tests on top of seq-ump-test.c in for-next.
Without patches 1-3, the new tests midi1_key_pressure,
ump_midi1_to_midi2_note and ump_midi2_to_midi1_note fail and the other
25 pass. With them, all 28 pass.
Prepared with Claude Code (Claude Opus 5.5). Tested on x86_64 under
qemu, on sound/for-next.
Best regards,
HyeongJun An
HyeongJun An (4):
ALSA: seq: ump: Don't turn a zero key pressure into a Note Off
ALSA: seq: ump: Convert a zero-velocity Note On to Note Off for MIDI
2.0
ALSA: seq: ump: Keep a soft MIDI 2.0 Note On from becoming a Note Off
selftests/alsa: Test the sequencer UMP Note On/Off conversion
sound/core/seq/seq_ump_convert.c | 10 +-
tools/testing/selftests/alsa/seq-ump-test.c | 187 +++++++++++++++++++-
2 files changed, 188 insertions(+), 9 deletions(-)
--
2.43.0
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH 1/4] ALSA: seq: ump: Don't turn a zero key pressure into a Note Off
2026-11-16 3:03 [PATCH 0/4] ALSA: seq: ump: Fix Note On/Off and key pressure conversion HyeongJun An
@ 2026-11-16 3:03 ` HyeongJun An
2026-11-16 3:03 ` [PATCH 2/4] ALSA: seq: ump: Convert a zero-velocity Note On to Note Off for MIDI 2.0 HyeongJun An
` (3 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: HyeongJun An @ 2026-11-16 3:03 UTC (permalink / raw)
To: Takashi Iwai
Cc: Jaroslav Kysela, Mark Brown, Shuah Khan, linux-sound,
linux-kselftest, linux-kernel, HyeongJun An, stable
The conversion of legacy events to UMP MIDI 1.0 uses
note_ev_to_ump_midi1() for KEYPRESS events as well as for note events.
The function turns any event with zero velocity into a Note Off. So a
Polyphonic Key Pressure of 0 reaches a UMP MIDI 1.0 client as a Note Off
and ends the note.
Only a Note On with zero velocity means Note Off. To fix this, check the
status, too.
The conversion to MIDI 2.0 already sends a zero key pressure as it is.
Fixes: e9e02819a98a ("ALSA: seq: Automatic conversion of UMP events")
Cc: stable@vger.kernel.org
Signed-off-by: HyeongJun An <sammiee5311@gmail.com>
---
sound/core/seq/seq_ump_convert.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/sound/core/seq/seq_ump_convert.c b/sound/core/seq/seq_ump_convert.c
index 62bdef6dca58..c248d87f7a83 100644
--- a/sound/core/seq/seq_ump_convert.c
+++ b/sound/core/seq/seq_ump_convert.c
@@ -630,7 +630,7 @@ static int note_ev_to_ump_midi1(const struct snd_seq_event *event,
union snd_ump_midi1_msg *data,
unsigned char status)
{
- if (!event->data.note.velocity)
+ if (status == UMP_MSG_STATUS_NOTE_ON && !event->data.note.velocity)
status = UMP_MSG_STATUS_NOTE_OFF;
data->note.status = status;
data->note.channel = event->data.note.channel & 0x0f;
--
2.43.0
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH 2/4] ALSA: seq: ump: Convert a zero-velocity Note On to Note Off for MIDI 2.0
2026-11-16 3:03 [PATCH 0/4] ALSA: seq: ump: Fix Note On/Off and key pressure conversion HyeongJun An
2026-11-16 3:03 ` [PATCH 1/4] ALSA: seq: ump: Don't turn a zero key pressure into a Note Off HyeongJun An
@ 2026-11-16 3:03 ` HyeongJun An
2026-11-16 3:03 ` [PATCH 3/4] ALSA: seq: ump: Keep a soft MIDI 2.0 Note On from becoming a Note Off HyeongJun An
` (2 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: HyeongJun An @ 2026-11-16 3:03 UTC (permalink / raw)
To: Takashi Iwai
Cc: Jaroslav Kysela, Mark Brown, Shuah Khan, linux-sound,
linux-kselftest, linux-kernel, HyeongJun An, stable
When a UMP MIDI 1.0 client sends to a MIDI 2.0 client,
cvt_ump_midi1_to_midi2() passes a Note On with zero velocity as a MIDI
2.0 Note On with zero velocity. The MIDI 1.0 sender meant a Note Off,
but the MIDI 2.0 receiver gets a Note On.
The other MIDI 1.0 to MIDI 2.0 conversions send a Note Off here:
note_ev_to_ump_midi2() for legacy sequencer events, and
cvt_legacy_cmd_to_ump() in the UMP core. Do the same.
Fixes: e9e02819a98a ("ALSA: seq: Automatic conversion of UMP events")
Cc: stable@vger.kernel.org
Signed-off-by: HyeongJun An <sammiee5311@gmail.com>
---
sound/core/seq/seq_ump_convert.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/sound/core/seq/seq_ump_convert.c b/sound/core/seq/seq_ump_convert.c
index c248d87f7a83..b35e3a50ec52 100644
--- a/sound/core/seq/seq_ump_convert.c
+++ b/sound/core/seq/seq_ump_convert.c
@@ -380,6 +380,10 @@ static int cvt_ump_midi1_to_midi2(struct snd_seq_client *dest,
switch (midi1->note.status) {
case UMP_MSG_STATUS_NOTE_ON:
case UMP_MSG_STATUS_NOTE_OFF:
+ /* MIDI 2.0 doesn't treat a note-on with 0 velocity as note-off */
+ if (midi1->note.status == UMP_MSG_STATUS_NOTE_ON &&
+ !midi1->note.velocity)
+ midi2->note.status = UMP_MSG_STATUS_NOTE_OFF;
midi2->note.note = midi1->note.note;
midi2->note.velocity = upscale_7_to_16bit(midi1->note.velocity);
break;
--
2.43.0
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH 3/4] ALSA: seq: ump: Keep a soft MIDI 2.0 Note On from becoming a Note Off
2026-11-16 3:03 [PATCH 0/4] ALSA: seq: ump: Fix Note On/Off and key pressure conversion HyeongJun An
2026-11-16 3:03 ` [PATCH 1/4] ALSA: seq: ump: Don't turn a zero key pressure into a Note Off HyeongJun An
2026-11-16 3:03 ` [PATCH 2/4] ALSA: seq: ump: Convert a zero-velocity Note On to Note Off for MIDI 2.0 HyeongJun An
@ 2026-11-16 3:03 ` HyeongJun An
2026-11-16 3:03 ` [PATCH 4/4] selftests/alsa: Test the sequencer UMP Note On/Off conversion HyeongJun An
2026-11-16 12:17 ` [PATCH 0/4] ALSA: seq: ump: Fix Note On/Off and key pressure conversion Takashi Iwai
4 siblings, 0 replies; 6+ messages in thread
From: HyeongJun An @ 2026-11-16 3:03 UTC (permalink / raw)
To: Takashi Iwai
Cc: Jaroslav Kysela, Mark Brown, Shuah Khan, linux-sound,
linux-kselftest, linux-kernel, HyeongJun An, stable
The cvt_ump_midi2_to_midi1() scales the 16-bit MIDI 2.0 velocity down to
7 bits. A Note On with a velocity below 0x200 becomes a MIDI 1.0 Note On
with zero velocity, which a MIDI 1.0 receiver takes as a Note Off.
Send velocity 1 then, as ump_midi2_to_note_ev() does for legacy
clients and cvt_ump_midi2_to_legacy() does in the UMP core.
Fixes: e9e02819a98a ("ALSA: seq: Automatic conversion of UMP events")
Cc: stable@vger.kernel.org
Signed-off-by: HyeongJun An <sammiee5311@gmail.com>
---
sound/core/seq/seq_ump_convert.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/sound/core/seq/seq_ump_convert.c b/sound/core/seq/seq_ump_convert.c
index b35e3a50ec52..b42a7ffbf31e 100644
--- a/sound/core/seq/seq_ump_convert.c
+++ b/sound/core/seq/seq_ump_convert.c
@@ -457,6 +457,10 @@ static int cvt_ump_midi2_to_midi1(struct snd_seq_client *dest,
case UMP_MSG_STATUS_NOTE_OFF:
midi1->note.note = midi2->note.note;
midi1->note.velocity = downscale_16_to_7bit(midi2->note.velocity);
+ /* a note-on with 0 velocity would be a note-off in MIDI 1.0 */
+ if (midi2->note.status == UMP_MSG_STATUS_NOTE_ON &&
+ !midi1->note.velocity)
+ midi1->note.velocity = 1;
break;
case UMP_MSG_STATUS_POLY_PRESSURE:
midi1->paf.note = midi2->paf.note;
--
2.43.0
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH 4/4] selftests/alsa: Test the sequencer UMP Note On/Off conversion
2026-11-16 3:03 [PATCH 0/4] ALSA: seq: ump: Fix Note On/Off and key pressure conversion HyeongJun An
` (2 preceding siblings ...)
2026-11-16 3:03 ` [PATCH 3/4] ALSA: seq: ump: Keep a soft MIDI 2.0 Note On from becoming a Note Off HyeongJun An
@ 2026-11-16 3:03 ` HyeongJun An
2026-11-16 12:17 ` [PATCH 0/4] ALSA: seq: ump: Fix Note On/Off and key pressure conversion Takashi Iwai
4 siblings, 0 replies; 6+ messages in thread
From: HyeongJun An @ 2026-11-16 3:03 UTC (permalink / raw)
To: Takashi Iwai
Cc: Jaroslav Kysela, Mark Brown, Shuah Khan, linux-sound,
linux-kselftest, linux-kernel, HyeongJun An
seq-ump-test only covers the RPN/NRPN conversion. Check that a Note On
with zero velocity becomes a Note Off and that a key pressure of zero
stays a key pressure, from a legacy client to UMP MIDI 1.0 and MIDI 2.0
clients. Also check Note On and Note Off between UMP MIDI 1.0 and MIDI
2.0 clients.
Without the three previous fixes, midi1_key_pressure,
ump_midi1_to_midi2_note and ump_midi2_to_midi1_note fail.
Signed-off-by: HyeongJun An <sammiee5311@gmail.com>
---
tools/testing/selftests/alsa/seq-ump-test.c | 187 +++++++++++++++++++-
1 file changed, 179 insertions(+), 8 deletions(-)
diff --git a/tools/testing/selftests/alsa/seq-ump-test.c b/tools/testing/selftests/alsa/seq-ump-test.c
index d1eca920cfc8..55bb7ea5b02f 100644
--- a/tools/testing/selftests/alsa/seq-ump-test.c
+++ b/tools/testing/selftests/alsa/seq-ump-test.c
@@ -1,11 +1,11 @@
// SPDX-License-Identifier: GPL-2.0
/*
- * Test the conversion of MIDI 1.0 sequencer events to UMP.
+ * Test the sequencer conversion of MIDI 1.0 events to UMP, and of UMP
+ * packets between MIDI 1.0 and MIDI 2.0 clients.
*
- * A legacy client sends CC and CONTROL14 events to a UMP client, and the
- * test checks the packets that the UMP client receives. The converter keeps
- * its RPN/NRPN state per destination port and channel, so each test uses
- * new clients.
+ * A client sends events to a UMP client, and the test checks the packets
+ * that the UMP client receives. The converter keeps its RPN/NRPN state
+ * per destination port and channel, so each test uses new clients.
*/
#include <errno.h>
#include <fcntl.h>
@@ -124,6 +124,48 @@ static int send_event(struct seq_client *src, struct seq_client *dst,
return 0;
}
+static int send_note(struct seq_client *src, struct seq_client *dst,
+ int type, int ch, int note, int velocity)
+{
+ struct snd_seq_event ev;
+
+ memset(&ev, 0, sizeof(ev));
+ ev.type = type;
+ ev.flags = SNDRV_SEQ_TIME_STAMP_TICK | SNDRV_SEQ_EVENT_LENGTH_FIXED;
+ ev.queue = SNDRV_SEQ_QUEUE_DIRECT;
+ ev.source.client = src->id;
+ ev.source.port = src->port;
+ ev.dest.client = dst->id;
+ ev.dest.port = dst->port;
+ ev.data.note.channel = ch;
+ ev.data.note.note = note;
+ ev.data.note.velocity = velocity;
+ if (write(src->fd, &ev, sizeof(ev)) != sizeof(ev))
+ return -errno;
+ return 0;
+}
+
+/* send a UMP packet from a UMP client */
+static int send_ump(struct seq_client *src, struct seq_client *dst,
+ unsigned int w0, unsigned int w1)
+{
+ struct snd_seq_ump_event ev;
+
+ memset(&ev, 0, sizeof(ev));
+ ev.flags = SNDRV_SEQ_TIME_STAMP_TICK | SNDRV_SEQ_EVENT_LENGTH_FIXED |
+ SNDRV_SEQ_EVENT_UMP;
+ ev.queue = SNDRV_SEQ_QUEUE_DIRECT;
+ ev.source.client = src->id;
+ ev.source.port = src->port;
+ ev.dest.client = dst->id;
+ ev.dest.port = dst->port;
+ ev.ump[0] = w0;
+ ev.ump[1] = w1;
+ if (write(src->fd, &ev, sizeof(ev)) != sizeof(ev))
+ return -errno;
+ return 0;
+}
+
/*
* Read what the UMP client received until it is quiet, and keep the
* packets of message type @mt. Returns the number of packets kept.
@@ -212,11 +254,11 @@ static int send_events(struct seq_client *src, struct seq_client *dst,
ARRAY_SIZE(ev))); \
} while (0)
-/* compare the RPN/NRPN packets received with the expected ones */
-#define EXPECT_PACKETS(...) do { \
+/* compare the packets of message type @mt with the expected ones */
+#define EXPECT_MT_PACKETS(mt, ...) do { \
const struct packet want[] = { __VA_ARGS__ }; \
struct packet got[MAX_PACKETS]; \
- int n = read_packets(&self->dst, 4, got); \
+ int n = read_packets(&self->dst, mt, got); \
int i; \
\
ASSERT_EQ((int)ARRAY_SIZE(want), n); \
@@ -226,12 +268,22 @@ static int send_events(struct seq_client *src, struct seq_client *dst,
} \
} while (0)
+/* compare the MIDI 2.0 channel voice packets received */
+#define EXPECT_PACKETS(...) EXPECT_MT_PACKETS(4, __VA_ARGS__)
+
#define EXPECT_NO_PACKETS() do { \
struct packet got[MAX_PACKETS]; \
\
EXPECT_EQ(0, read_packets(&self->dst, 4, got)); \
} while (0)
+#define SEND_NOTE(type, ch, note, velocity) \
+ ASSERT_EQ(0, send_note(&self->src, &self->dst, \
+ SNDRV_SEQ_EVENT_##type, ch, note, velocity))
+
+#define SEND_UMP(w0, w1) \
+ ASSERT_EQ(0, send_ump(&self->src, &self->dst, w0, w1))
+
/* RPN 0/1 with Data Entry */
TEST_F(seq_ump, rpn)
{
@@ -416,6 +468,20 @@ TEST_F(seq_ump, c14_null_nrpn_keeps_rpn_number)
EXPECT_PACKETS({ RPN(0, 0, 1), data14(0x10, 0) });
}
+/* a note-on with zero velocity is a note-off */
+TEST_F(seq_ump, note_on_zero_velocity)
+{
+ SEND_NOTE(NOTEON, 1, 60, 0);
+ EXPECT_PACKETS({ 0x40813c00, 0 });
+}
+
+/* a key pressure of zero is a key pressure, not a note-off */
+TEST_F(seq_ump, key_pressure)
+{
+ SEND_NOTE(KEYPRESS, 2, 64, 0);
+ EXPECT_PACKETS({ 0x40a24000, 0 });
+}
+
/* a UMP MIDI 1.0 client gets the CCs as they are, RPN selection included */
TEST(midi1_passthrough)
{
@@ -449,4 +515,109 @@ TEST(midi1_passthrough)
close_client(&dst);
}
+/* a pair of clients whose MIDI versions each test chooses */
+FIXTURE(seq_pair) {
+ struct seq_client src;
+ struct seq_client dst;
+};
+
+FIXTURE_SETUP(seq_pair)
+{
+ self->src.fd = -1;
+ self->dst.fd = -1;
+}
+
+FIXTURE_TEARDOWN(seq_pair)
+{
+ close_client(&self->src);
+ close_client(&self->dst);
+}
+
+static int open_pair(struct seq_client *dst, int dst_version,
+ struct seq_client *src, int src_version, int *skip)
+{
+ int err;
+
+ err = open_client(dst, dst_version, CAP_IN, skip);
+ if (!err)
+ err = open_client(src, src_version, CAP_OUT, skip);
+ return err;
+}
+
+/* to UMP MIDI 1.0, a key pressure of zero is a key pressure */
+TEST_F(seq_pair, midi1_key_pressure)
+{
+ int skip = 0;
+ int err;
+
+ err = open_pair(&self->dst, SNDRV_SEQ_CLIENT_UMP_MIDI_1_0,
+ &self->src, 0, &skip);
+ if (skip)
+ SKIP(return, "UMP sequencer client unavailable: %s",
+ strerror(-err));
+ ASSERT_EQ(0, err);
+
+ SEND_NOTE(KEYPRESS, 2, 64, 0x50);
+ SEND_NOTE(KEYPRESS, 2, 64, 0);
+ EXPECT_MT_PACKETS(2, { 0x20a24050, 0 }, { 0x20a24000, 0 });
+}
+
+/* to UMP MIDI 1.0, a note-on with zero velocity is a note-off */
+TEST_F(seq_pair, midi1_note_on_zero_velocity)
+{
+ int skip = 0;
+ int err;
+
+ err = open_pair(&self->dst, SNDRV_SEQ_CLIENT_UMP_MIDI_1_0,
+ &self->src, 0, &skip);
+ if (skip)
+ SKIP(return, "UMP sequencer client unavailable: %s",
+ strerror(-err));
+ ASSERT_EQ(0, err);
+
+ SEND_NOTE(NOTEON, 1, 60, 0);
+ EXPECT_MT_PACKETS(2, { 0x20813c00, 0 });
+}
+
+/* UMP MIDI 1.0 to MIDI 2.0: a note-on with zero velocity is a note-off */
+TEST_F(seq_pair, ump_midi1_to_midi2_note)
+{
+ int skip = 0;
+ int err;
+
+ err = open_pair(&self->dst, SNDRV_SEQ_CLIENT_UMP_MIDI_2_0,
+ &self->src, SNDRV_SEQ_CLIENT_UMP_MIDI_1_0, &skip);
+ if (skip)
+ SKIP(return, "UMP sequencer client unavailable: %s",
+ strerror(-err));
+ ASSERT_EQ(0, err);
+
+ SEND_UMP(0x20913c00, 0);
+ EXPECT_PACKETS({ 0x40813c00, 0 });
+}
+
+/*
+ * UMP MIDI 2.0 to MIDI 1.0: a note-on whose velocity scales down to zero
+ * stays a note-on with velocity 1, and a note-off stays a note-off
+ */
+TEST_F(seq_pair, ump_midi2_to_midi1_note)
+{
+ int skip = 0;
+ int err;
+
+ err = open_pair(&self->dst, SNDRV_SEQ_CLIENT_UMP_MIDI_1_0,
+ &self->src, SNDRV_SEQ_CLIENT_UMP_MIDI_2_0, &skip);
+ if (skip)
+ SKIP(return, "UMP sequencer client unavailable: %s",
+ strerror(-err));
+ ASSERT_EQ(0, err);
+
+ SEND_UMP(0x40913c00, 0xc8000000);
+ SEND_UMP(0x40913c00, 0x01000000);
+ SEND_UMP(0x40913c00, 0);
+ SEND_UMP(0x40813c00, 0);
+ EXPECT_MT_PACKETS(2, { 0x20913c64, 0 }, { 0x20913c01, 0 },
+ { 0x20913c01, 0 }, { 0x20813c00, 0 });
+}
+
TEST_HARNESS_MAIN
--
2.43.0
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH 0/4] ALSA: seq: ump: Fix Note On/Off and key pressure conversion
2026-11-16 3:03 [PATCH 0/4] ALSA: seq: ump: Fix Note On/Off and key pressure conversion HyeongJun An
` (3 preceding siblings ...)
2026-11-16 3:03 ` [PATCH 4/4] selftests/alsa: Test the sequencer UMP Note On/Off conversion HyeongJun An
@ 2026-11-16 12:17 ` Takashi Iwai
4 siblings, 0 replies; 6+ messages in thread
From: Takashi Iwai @ 2026-11-16 12:17 UTC (permalink / raw)
To: HyeongJun An
Cc: Takashi Iwai, Jaroslav Kysela, Mark Brown, Shuah Khan,
linux-sound, linux-kselftest, linux-kernel
On Mon, 16 Nov 2026 04:03:33 +0100,
HyeongJun An wrote:
>
> Hello Takashi,
>
> While extending seq-ump-test, I found three conversion bugs in
> seq_ump_convert.c. In each case the UMP core or the other sequencer
> path already does the right thing, and the fix follows it.
>
> Patches 1-3 are the fixes. They also apply to for-linus. Patch 4 adds
> the tests on top of seq-ump-test.c in for-next.
>
> Without patches 1-3, the new tests midi1_key_pressure,
> ump_midi1_to_midi2_note and ump_midi2_to_midi1_note fail and the other
> 25 pass. With them, all 28 pass.
>
> Prepared with Claude Code (Claude Opus 5.5). Tested on x86_64 under
> qemu, on sound/for-next.
>
> Best regards,
> HyeongJun An
>
> HyeongJun An (4):
> ALSA: seq: ump: Don't turn a zero key pressure into a Note Off
> ALSA: seq: ump: Convert a zero-velocity Note On to Note Off for MIDI
> 2.0
> ALSA: seq: ump: Keep a soft MIDI 2.0 Note On from becoming a Note Off
> selftests/alsa: Test the sequencer UMP Note On/Off conversion
Applied all four patches now to for-next branch. Thanks.
Takashi
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2026-11-16 12:17 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-11-16 3:03 [PATCH 0/4] ALSA: seq: ump: Fix Note On/Off and key pressure conversion HyeongJun An
2026-11-16 3:03 ` [PATCH 1/4] ALSA: seq: ump: Don't turn a zero key pressure into a Note Off HyeongJun An
2026-11-16 3:03 ` [PATCH 2/4] ALSA: seq: ump: Convert a zero-velocity Note On to Note Off for MIDI 2.0 HyeongJun An
2026-11-16 3:03 ` [PATCH 3/4] ALSA: seq: ump: Keep a soft MIDI 2.0 Note On from becoming a Note Off HyeongJun An
2026-11-16 3:03 ` [PATCH 4/4] selftests/alsa: Test the sequencer UMP Note On/Off conversion HyeongJun An
2026-11-16 12:17 ` [PATCH 0/4] ALSA: seq: ump: Fix Note On/Off and key pressure conversion Takashi Iwai
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®