tm1300 NXP Semiconductors, tm1300 Datasheet - Page 127

• Manufacturer Part Number: tm1300
• Description: Tm-1300 Media Processor
• Manufacturer: NXP Semiconductors

Philips Semiconductors
Note: In this mode, SYNC_MASTER must be set to en-
sure correct operation of VO_IO1 and VO_IO2 as out-
puts.
When each buffer has been transferred, the correspond-
ing buffer-empty bit is set in the status register, and the
DSPCPU is interrupted if the buffer-empty interrupt is en-
abled. To maintain continuous transfer of data, the
DSPCPU supplies new pointers for the next data buffer
following each buffer-empty interrupt. If the DSPCPU
does not supply new pointers before the next field, the
URUN bit is set, and the EVO uses the same pointer val-
ues until they are updated.
When
EVO_ENABLE = 1 and SYNC_STREAMING = 1, the
VO_IO2 signal indicates a data-valid condition. This sig-
nal is asserted when the EVO starts outputing valid data
(that is, data-streaming mode is enabled and video out-
put is running) and is de-asserted when data-streaming
mode is disabled. The VO_IO1 signal generates a pulse
one VO_CLK cycle before the first valid data is sent. See
Section 7.11
Message-Passing Mode . In message-passing mode
data is stored in SDRAM in one buffer.
Note: In this mode, SYNC_MASTER must be set to en-
sure correct operation of VO_IO1 and VO_IO2 as out-
puts.
When
VO_CTL.VO_ENABLE, the EVO sends a Start condition
on VO_IO1. When the EVO has transferred the contents
of the buffer, it sends an End condition on VO_IO2 as
shown in
rupts the DSPCPU. The EVO stops, and no further oper-
ation takes place until the DSPCPU sets VO_ENABLE
again to start another message, or until the DSCPU ini-
tiates other EVO operation. See
signal details.
7.17.3
The EVO has five interrupt conditions defined by bits in
the
BFR2_EMPTY, HBE, URUN, and YTR. Each of these
conditions has a corresponding interrupt enable flag and
interrupt acknowledge bit in the VO_CTL register.
The EVO asserts a SOURCE 10 interrupt request to the
TM1300 vectored interrupt controller as long as one or
more enabled events is asserted.
Note: The interrupt controller should always be pro-
grammed such that the EVO interrupt operates in level-
triggered mode. This ensures that no EVO events can be
lost to the interrupt handler. Refer to
and NMI (Maskable and Non-Maskable Interrupts),”
a description of setting level-triggered mode, as well as
for recommendations on writing interrupt handlers.
The BFR1_EMPTY, BFR2_EMPTY and YTR status
flags indicate to the DSPCPU that a buffer has been
emptied or that the Y threshold has been reached.
The buffer-underrun (URUN) status flag indicates that
the DSPCPU did not acknowledge a BFR1_EMPTY or
VO_STATUS
message
data-streaming
Interrupts and Error Conditions
Figure
for timing signal details.
7-18, sets BFR1_EMPTY, and inter-
passing
register:
mode
is
Section 7.11
started
is
Section 3.5.3, “INT
BFR1_EMPTY,
enabled
by
for timing
setting
and
for
BFR2_EMPTY interrupt before the EVO required the
next buffer. In this case, the EVO uses the old address
pointer value and continues image or data transfer.
When the DSPCPU updates the pointer, the new pointer
value will be used at the start of the next frame or buffer
transfer. Therefore, the URUN flag can be interpreted as
indicating to the DSPCPU that the EVO is using its old
pointer values because it did not receive the new ones in
time.
Note: The actual buffer pointer write operation to the
MMIO registers is not seen by the hardware—only writ-
ing a ’1’ to the appropriate BFR1_ACK or BFR2_ACK
bits signals buffer availability.
The Hardware Bandwidth Error (HBE) flag indicates that
the EVO did not get data from SDRAM via the TM1300’s
internal data highway in time to continue data transfer or
video refresh. Data or video refresh will continue using
whatever data is in the EVO internal data buffers. The
address counter for the failing buffer(s) will continue to
count, and the EVO will continue to request data from the
SDRAM over the highway.
The EVO is a read-only device, transferring data from
SDRAM to the EVO output port. Unlike Video In, the
EVO does not modify SDRAM data. URUN and HBE are
the only EVO error conditions that can arise. In the case
of URUN or HBE, a scrambled image may be temporarily
displayed or incorrect data may be temporarily sent. The
EVO can cause no other system hardware error condi-
tions.
Even changing operating modes can not cause system
hardware error conditions to arise. For example, chang-
ing the MODE bits, the OL_EN and format bits, or the
LTL_END bit while the EVO is running may cause wrong
data to be displayed or transferred. However, the EVO
does not detect this or stop for it.
In normal operation, the user should not change the
mode or transfer-control bits while the EVO is enabled.
The EVO should be disabled before changing bits such
as the MODE bits, the OL_EN bit, or the LTL_END bit.
However if these bits are changed while the EVO is run-
ning, they will take effect at the beginning of the next field
or buffer.
7.17.4
In order to avoid Hardware Bandwidth Error (HBE) con-
ditions, the internal highway bus arbiter (see
“Arbiter”) must be programmed according to the latency
requirements of the EVO unit described in this section. In
the following discussion, it is assumed that data for video
lines (in Y, U, V and overlay planar memory format) is
stored in memory aligned on 64-byte boundaries. In oth-
er words, it means that the {OL,Y,U,V}_OFFSET fields
are multiples of 64 bytes. Otherwise internal EVO arbitra-
tion for OL, Y, U and V requests will be different than de-
scribed here, and the following latencies would not be
guaranteed. The EVO uses internal 64-byte buffers.
1. Latency requirements for the EVO in image mode
PRODUCT SPECIFICATION
4:2:2 or 4:2:0 co-sited or interspersed without upscal-
ing and with overlay disabled is expressed as follows.
Latency and Bandwidth Requirements
Enhanced Video Out
Chapter 20,
7-23