dm00068628

STM32F42xx and STM32F43xx
Errata sheet
STM32F427/437 and STM32F429/439 line limitations
Silicon identification
This errata sheet applies to the revision A, Y, 1 and 3 of STMicroelectronics
STM32F427/437 and STM32F429/439 microcontroller lines.
The STM32F42xx and STM32F43xx devices feature an ARM® 32-bit Cortex®-M4 core with
FPU, for which an errata notice is also available (see Section 1 for details).
The full list of part numbers is shown in Table 2. The products are identifiable as shown in
Table 1:
•
by the revision code marked below the order code on the device package
•
by the last three digits of the Internal order code printed on the box label
Table 1. Device identification(1)
Revision code marked on device(2)
Order code
STM32F427xx, STM32F429xx
“A”, “Y”, “1”, “3”
STM32F437xx, STM32F439xx
1. The REV_ID bits in the DBGMCU_IDCODE register show the revision code of the device (see the RM0090
STM32F4xx reference manual for details on how to find the revision code).
2. Refer to the device datasheets for details on how to identify the revision code and the date code on the
different packages.
Table 2. Device summary
Reference
October 2014
Part number
STM32F427xx
STM32F427VG, STM32F427ZG, STM32F427IG, STM32F427AG,
STM32F427VI, STM32F427ZI, STM32F427II, STM32F427AI
STM32F437xx
STM32F437VG, STM32F437ZG, STM32F437IG, STM32F437VI, STM32F437ZI,
STM32F437II, STM32F437AI
STM32F429xx
STM32F429VG, STM32F429ZG, STM32F429IG, STM32F429VI, STM32F429ZI,
STM32F429II, STM32F429AI, STM32F429AG, STM32F429BG, STM32F429BI,
STM32F429NI, STM32F429NG, STM32F429VE, STM32F429ZE,
STM32F429IE, STM32F429BE, STM32F429NE
STM32F439xx
STM32F439VI, STM32F439VG, STM32F439ZG, STM32F439ZI, STM32F439IG,
STM32F439II, STM32F439AI, STM32F439BG, STM32F439BI, STM32F439NI,
STM32F439NG
DocID023833 Rev 9
1/34
www.st.com
1
Contents
STM32F42xx and STM32F43xx
Contents
1
2
ARM 32-bit Cortex-M4 with FPU limitations . . . . . . . . . . . . . . . . . . . . . . 6
1.1
Cortex-M4 interrupted loads to stack pointer can cause
erroneous behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2
VDIV or VSQRT instructions might not complete correctly
when very short ISRs are used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
STM32F42xx and STM32F43xx silicon limitations . . . . . . . . . . . . . . . . . 8
2.1
2.2
System limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11
2.1.1
Debugging Stop mode and system tick timer . . . . . . . . . . . . . . . . . . . . 11
2.1.2
Debugging Stop mode with WFE entry . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.3
Wakeup sequence from Standby mode when using more than
one wakeup source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.4
Full JTAG configuration without NJTRST pin cannot be used . . . . . . . . 12
2.1.5
MPU attribute to RTC and IWDG registers could be managed
incorrectly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.1.6
Delay after an RCC peripheral clock enabling . . . . . . . . . . . . . . . . . . . . 12
2.1.7
Internal noise impacting the ADC accuracy . . . . . . . . . . . . . . . . . . . . . . 13
2.1.8
Over-drive and Under-drive modes unavailability . . . . . . . . . . . . . . . . . 13
2.1.9
Operating voltage extension down to 1.7 V
in the whole temperature range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1.10
PA12 GPIO limitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
IWDG peripheral limitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.1
2.3
2.4
I2C peripheral limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3.1
SMBus standard not fully supported . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3.2
Start cannot be generated after a misplaced Stop . . . . . . . . . . . . . . . . . 15
2.3.3
Mismatch on the “Setup time for a repeated Start condition” timing
parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3.4
Data valid time (tVD;DAT) violated without the OVR flag being set . . . . . 16
2.3.5
Both SDA and SCL maximum rise time (tr) violated when VDD_I2C bus
higher than ((VDD+0.3) / 0.7) V . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
I2S peripheral limitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4.1
2.5
In I2S slave mode, WS level must be set by the external master
when enabling the I2S . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
USART peripheral limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5.1
2/34
RVU and PVU flags are not reset in STOP mode . . . . . . . . . . . . . . . . . 14
Idle frame is not detected if receiver clock speed is deviated . . . . . . . . 17
DocID023833 Rev 9
STM32F42xx and STM32F43xx
2.6
2.5.2
In full duplex mode, the Parity Error (PE) flag can be cleared by
writing to the data register . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5.3
Parity Error (PE) flag is not set when receiving in Mute mode
using address mark detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.5.4
Break frame is transmitted regardless of nCTS input line status . . . . . . 18
2.5.5
nRTS signal abnormally driven low after a protocol violation . . . . . . . . 18
bxCAN limitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.6.1
2.7
2.8
2.9
Contents
bxCAN time triggered communication mode not supported . . . . . . . . . 19
OTG_FS peripheral limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.7.1
Data in RxFIFO is overwritten when all channels are disabled
simultaneously . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.7.2
OTG host blocks the receive channel when receiving IN packets and no
TxFIFO is configured . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.7.3
Host channel-halted interrupt not generated when the channel is
disabled . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.7.4
Error in software-read OTG_FS_DCFG register values . . . . . . . . . . . . 20
Ethernet peripheral limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.8.1
Incorrect layer 3 (L3) checksum is inserted in transmitted IPv6 packets
without TCP, UDP or ICMP payloads . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.8.2
The Ethernet MAC processes invalid extension headers in the received
IPv6 frames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.8.3
MAC stuck in the Idle state on receiving the TxFIFO flush command
exactly 1 clock cycle after a transmission completes . . . . . . . . . . . . . . . 21
2.8.4
Transmit frame data corruption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.8.5
Successive write operations to the same register might not be fully
taken into account . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
FMC peripheral limitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.9.1
Dummy read cycles inserted when reading synchronous memories . . . 25
2.9.2
FMC synchronous mode and NWAIT signal disabled . . . . . . . . . . . . . . 25
2.9.3
Read access to a non-initialized FMC_SDRAM bank . . . . . . . . . . . . . . 25
2.9.4
Corruption of data read from the FMC . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.9.5
Interruption of CPU read burst access to an end of SDRAM row . . . . . 26
2.9.6
FMC NOR/PSRAM controller: asynchronous read access on bank 2 to 4
returns wrong data when bank 1 is in synchronous mode
(BURSTEN bit is set) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.9.7
FMC dynamic and static bank switching . . . . . . . . . . . . . . . . . . . . . . . . 27
2.9.8
NAND/PCCard transaction and Wait timing . . . . . . . . . . . . . . . . . . . . . . 27
2.9.9
Data corruption during burst read from FMC synchronous memory . . . 27
2.9.10
Missed burst write transaction on multiplexed PSRAM . . . . . . . . . . . . . 28
2.9.11
FMC NOR/PSRAM controller write protocol violation . . . . . . . . . . . . . . 28
DocID023833 Rev 9
3/34
4
Contents
STM32F42xx and STM32F43xx
2.9.12
2.10
2.11
SDIO peripheral limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.10.1
SDIO HW flow control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.10.2
Wrong CCRCFAIL status after a response without CRC is received . . . 29
2.10.3
Data corruption in SDIO clock dephasing (NEGEDGE) mode . . . . . . . . 29
2.10.4
CE-ATA multiple write command and card busy signal management . . 29
2.10.5
No underrun detection with wrong data transmission . . . . . . . . . . . . . . 30
ADC peripheral limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.11.1
2.12
3
4/34
FMC NOR/PSRAM controller bank switch with different
BUSTURN durations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
ADC sequencer modification during conversion . . . . . . . . . . . . . . . . . . 30
DAC peripheral limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.12.1
DMA underrun flag management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.12.2
DMA request not automatically cleared by DMAEN=0 . . . . . . . . . . . . . 31
Revision history . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
DocID023833 Rev 9
STM32F42xx and STM32F43xx
List of tables
List of tables
Table 1.
Table 2.
Table 3.
Table 4.
Table 5.
Table 6.
Device identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Device summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Cortex-M4 core limitations and impact on microcontroller behavior . . . . . . . . . . . . . . . . . . . 6
Summary of silicon limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Impacted registers and bits. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Document revision history . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
DocID023833 Rev 9
5/34
5
ARM 32-bit Cortex-M4 with FPU limitations
1
STM32F42xx and STM32F43xx
ARM 32-bit Cortex-M4 with FPU limitations
An errata notice of the STM32F42xx and STM32F43xx core is available from
http://infocenter.arm.com.
All the described limitations are minor and related to the revision r0p1-v1 of the Cortex-M4
core. Table 3 summarizes these limitations and their implications on the behavior of
STM32F42xx and STM32F43xx devices.
Table 3. Cortex-M4 core limitations and impact on microcontroller behavior
1.1
ARM ID
ARM
category
752770
Cat B
Interrupted loads to SP can cause erroneous
behavior
Minor
776924
Cat B
VDIV or VSQRT instructions might not complete
correctly when very short ISRs are used
Minor
ARM summary of errata
Impact on STM32F42xx
and STM32F43xx
Cortex-M4 interrupted loads to stack pointer can cause
erroneous behavior
Description
An interrupt occurring during the data-phase of a single word load to the stack pointer
(SP/R13) can cause an erroneous behavior of the device. In addition, returning from the
interrupt results in the load instruction being executed an additional time.
For all the instructions performing an update of the base register, the base register is
erroneously updated on each execution, resulting in the stack pointer being loaded from an
incorrect memory location.
The instructions affected by this limitation are the following:
•
LDR SP, [Rn],#imm
•
LDR SP, [Rn,#imm]!
•
LDR SP, [Rn,#imm]
•
LDR SP, [Rn]
•
LDR SP, [Rn,Rm]
Workaround
As of today, no compiler generates these particular instructions. This limitation can only
occur with hand-written assembly code.
Both limitations can be solved by replacing the direct load to the stack pointer by an
intermediate load to a general-purpose register followed by a move to the stack pointer.
Example:
Replace LDR SP, [R0] by
LDR R2,[R0]
MOV SP,R2
6/34
DocID023833 Rev 9
STM32F42xx and STM32F43xx
1.2
ARM 32-bit Cortex-M4 with FPU limitations
VDIV or VSQRT instructions might not complete correctly
when very short ISRs are used
Description
On Cortex-M4 with FPU core, 14 cycles are required to execute a VDIV or VSQRT
instruction.
This limitation is present when the following conditions are met:
•
A VDIV or VSQRT is executed
•
The destination register for VDIV or VSQRT is one of s0 - s15
•
An interrupt occurs and is taken
•
The ISR being executed does not contain a floating point instruction
•
14 cycles after the VDIV or VSQRT is executed, an interrupt return is executed
In this case, if there are only one or two instructions inside the interrupt service routine, then
the VDIV or VQSRT instruction does not complete correctly and the register bank and
FPSCR are not updated, meaning that these registers hold incorrect out-of-date data.
Workaround
Two workarounds are applicable:
•
Disable lazy context save of floating point state by clearing LSPEN to 0 (bit 30 of the
FPCCR at address 0xE000EF34).
•
Ensure that every ISR contains more than 2 instructions in addition to the exception
return instruction.
DocID023833 Rev 9
7/34
33
STM32F42xx and STM32F43xx silicon limitations
2
STM32F42xx and STM32F43xx
STM32F42xx and STM32F43xx silicon limitations
Table 4 gives quick references to all documented limitations.
Legend for Table 4: A = workaround available; N = no workaround available; P = partial
workaround available, ‘-’ and grayed = fixed.
Table 4. Summary of silicon limitations
Links to silicon limitations
Revision Revision Revision Revision
A
Y
1
3
Section 2.1.1: Debugging Stop mode and system
tick timer
A
A
A
A
Section 2.1.2: Debugging Stop mode with WFE
entry
A
A
A
A
Section 2.1.3: Wakeup sequence from Standby
mode when using more than one wakeup source
A
A
A
A
Section 2.1.4: Full JTAG configuration without
NJTRST pin cannot be used
A
A
A
A
A
A
A
A
A
A
A
A
Section 2.1.7: Internal noise impacting the ADC
accuracy
A
A
A
A
Section 2.1.8: Over-drive and Under-drive modes
unavailability
N
-
-
-
Section 2.1.9: Operating voltage extension down to
1.7 V in the whole temperature range
N
N
-
-
Section 2.1.10: PA12 GPIO limitation
A
A
A
-
Section 2.2: IWDG
Section 2.2.1: RVU and PVU flags are not reset in
peripheral
STOP mode
limitation
A
A
A
A
Section 2.3.1: SMBus standard not fully supported
A
A
A
A
Section 2.3.2: Start cannot be generated after a
misplaced Stop
A
A
A
A
Section 2.3.3: Mismatch on the “Setup time for a
repeated Start condition” timing parameter
A
A
A
A
Section 2.3.4: Data valid time (tVD;DAT) violated
without the OVR flag being set
A
A
A
A
Section 2.3.5: Both SDA and SCL maximum rise
time (tr) violated when VDD_I2C bus higher than
((VDD+0.3) / 0.7) V
A
A
A
A
Section 2.4.1: In I2S slave mode, WS level must
be set by the external master when enabling the
I2S
A
A
A
A
Section 2.1.5: MPU attribute to RTC and IWDG
Section 2.1:
registers could be managed incorrectly
System limitations
Section 2.1.6: Delay after an RCC peripheral clock
enabling
Section 2.3: I2C
peripheral
limitations
Section 2.4: I2S
peripheral
limitation
8/34
DocID023833 Rev 9
STM32F42xx and STM32F43xx
STM32F42xx and STM32F43xx silicon limitations
Table 4. Summary of silicon limitations (continued)
Links to silicon limitations
Revision Revision Revision Revision
A
Y
1
3
Section 2.5.1: Idle frame is not detected if receiver
clock speed is deviated
N
N
N
N
Section 2.5.2: In full duplex mode, the Parity Error
(PE) flag can be cleared by writing to the data
register
A
A
A
A
Section 2.5:
Section 2.5.3: Parity Error (PE) flag is not set when
USART peripheral
receiving in Mute mode using address mark
limitations
detection
N
N
N
N
Section 2.5.4: Break frame is transmitted
regardless of nCTS input line status
N
N
N
N
Section 2.5.5: nRTS signal abnormally driven low
after a protocol violation
A
A
A
A
Section 2.6.1: bxCAN time triggered
communication mode not supported
N
N
N
N
Section 2.7.1: Data in RxFIFO is overwritten when
all channels are disabled simultaneously
A
A
A
A
Section 2.7.2: OTG host blocks the receive
channel when receiving IN packets and no TxFIFO
is configured
A
A
A
A
Section 2.7.3: Host channel-halted interrupt not
generated when the channel is disabled
A
A
A
A
Section 2.7.4: Error in software-read
OTG_FS_DCFG register values
A
A
A
A
Section 2.8.1: Incorrect layer 3 (L3) checksum is
inserted in transmitted IPv6 packets without TCP,
UDP or ICMP payloads
A
A
A
A
Section 2.8.2: The Ethernet MAC processes invalid
extension headers in the received IPv6 frames
N
N
N
N
Section 2.8.3: MAC stuck in the Idle state on
receiving the TxFIFO flush command exactly 1
clock cycle after a transmission completes
A
A
A
A
Section 2.8.4: Transmit frame data corruption
A
A
A
A
Section 2.8.5: Successive write operations to the
same register might not be fully taken into account
A
A
A
A
Section 2.6:
bxCAN limitation
Section 2.7:
OTG_FS
peripheral
limitations
Section 2.8:
Ethernet
peripheral
limitations
DocID023833 Rev 9
9/34
33
STM32F42xx and STM32F43xx silicon limitations
STM32F42xx and STM32F43xx
Table 4. Summary of silicon limitations (continued)
Links to silicon limitations
Section 2.9: FMC
peripheral
limitation
Section 2.10:
SDIO peripheral
limitations
Section 2.9.1: Dummy read cycles inserted when
reading synchronous memories
N
N
N
N
Section 2.9.2: FMC synchronous mode and
NWAIT signal disabled
A
A
A
-
Section 2.9.3: Read access to a non-initialized
FMC_SDRAM bank
P
P
P
-
Section 2.9.4: Corruption of data read from the
FMC
A
-
-
-
Section 2.9.5: Interruption of CPU read burst
access to an end of SDRAM row
A
A
A
-
Section 2.9.6: FMC NOR/PSRAM controller:
asynchronous read access on bank 2 to 4 returns
wrong data when bank 1 is in synchronous mode
(BURSTEN bit is set)
A
A
-
-
Section 2.9.7: FMC dynamic and static bank
switching
A
A
A
-
Section 2.9.8: NAND/PCCard transaction and Wait
timing
A
A
A
A
Section 2.9.9: Data corruption during burst read
from FMC synchronous memory
A
A
A
A
Section 2.9.10: Missed burst write transaction on
multiplexed PSRAM
-
-
A
A
Section 2.9.11: FMC NOR/PSRAM controller write
protocol violation
N
N
N
-
Section 2.9.12: FMC NOR/PSRAM controller bank
switch with different BUSTURN durations
A
A
A
A
Section 2.10.1: SDIO HW flow control
N
N
N
N
Section 2.10.2: Wrong CCRCFAIL status after a
response without CRC is received
A
A
A
A
Section 2.10.3: Data corruption in SDIO clock
dephasing (NEGEDGE) mode
N
N
N
N
Section 2.10.4: CE-ATA multiple write command
and card busy signal management
A
A
A
A
Section 2.10.5: No underrun detection with wrong
data transmission
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
Section 2.11: ADC
Section 2.11.1: ADC sequencer modification during
peripheral
conversion
limitations
Section 2.12: DAC Section 2.12.1: DMA underrun flag management
peripheral
Section 2.12.2: DMA request not automatically
limitations
cleared by DMAEN=0
10/34
Revision Revision Revision Revision
A
Y
1
3
DocID023833 Rev 9
STM32F42xx and STM32F43xx
STM32F42xx and STM32F43xx silicon limitations
2.1
System limitations
2.1.1
Debugging Stop mode and system tick timer
Description
If the system tick timer interrupt is enabled during the Stop mode debug (DBG_STOP bit set
in the DBGMCU_CR register), it will wake up the system from Stop mode.
Workaround
To debug the Stop mode, disable the system tick timer interrupt.
2.1.2
Debugging Stop mode with WFE entry
Description
When the Stop debug mode is enabled (DBG_STOP bit set in the DBGMCU_CR register),
this allows software debugging during Stop mode.
However, if the application software uses the WFE instruction to enter Stop mode, after
wakeup some instructions could be missed if the WFE is followed by sequential instructions.
This affects only Stop debug mode with WFE entry.
Workaround
To debug Stop mode with WFE entry, the WFE instruction must be inside a dedicated
function with 1 instruction (NOP) between the execution of the WFE and the Bx LR.
Example:
__asm void _WFE(void) {
WFE
NOP
BX lr }
2.1.3
Wakeup sequence from Standby mode when using more than
one wakeup source
Description
The various wakeup sources are logically OR-ed in front of the rising-edge detector which
generates the wakeup flag (WUF). The WUF needs to be cleared prior to Standby mode
entry, otherwise the MCU wakes up immediately.
If one of the configured wakeup sources is kept high during the clearing of the WUF (by
setting the CWUF bit), it may mask further wakeup events on the input of the edge detector.
As a consequence, the MCU might not be able to wake up from Standby mode.
DocID023833 Rev 9
11/34
33
STM32F42xx and STM32F43xx silicon limitations
STM32F42xx and STM32F43xx
Workaround
To avoid this problem, the following sequence should be applied before entering
Standby mode:
•
Disable all used wakeup sources,
•
Clear all related wakeup flags,
•
Re-enable all used wakeup sources,
•
Enter Standby mode
Note:
Be aware that, when applying this workaround, if one of the wakeup sources is still kept
high, the MCU enters Standby mode but then it wakes up immediately generating a power
reset.
2.1.4
Full JTAG configuration without NJTRST pin cannot be used
Description
When using the JTAG debug port in debug mode, the connection with the debugger is lost if
the NJTRST pin (PB4) is used as a GPIO. Only the 4-wire JTAG port configuration is
impacted.
Workaround
Use the SWD debug port instead of the full 4-wire JTAG port.
2.1.5
MPU attribute to RTC and IWDG registers could be managed
incorrectly
Description
If the MPU is used and the non bufferable attribute is set to the RTC or IWDG memory map
region, the CPU access to the RTC or IWDG registers could be treated as bufferable,
provided that there is no APB prescaler configured (AHB/APB prescaler is equal to 1).
Workaround
If the non bufferable attribute is required for these registers, the software could perform a
read after the write to guaranty the completion of the write access.
2.1.6
Delay after an RCC peripheral clock enabling
Description
A delay between an RCC peripheral clock enable and the effective peripheral enabling
should be taken into account in order to manage the peripheral read/write to registers.
This delay depends on the peripheral’s mapping:
12/34
•
If the peripheral is mapped on AHB: the delay should be equal to 2 AHB cycles.
•
If the peripheral is mapped on APB: the delay should be equal to 1 + (AHB/APB
prescaler) cycles.
DocID023833 Rev 9
STM32F42xx and STM32F43xx
STM32F42xx and STM32F43xx silicon limitations
Workarounds
2.1.7
1.
Use the DSB instruction to stall the Cortex-M4 CPU pipeline until the instruction is
completed.
2.
Insert “n” NOPs between the RCC enable bit write and the peripheral register writes
(n = 2 for AHB peripherals, n = 1 + AHB/APB prescaler in case of APB peripherals).
Internal noise impacting the ADC accuracy
Description
An internal noise generated on VDD supplies and propagated internally may impact the ADC
accuracy.
This noise is always active whatever the power mode of the MCU (RUN or Sleep).
Workarounds
To adapt the accuracy level to the application requirements, set one of the following options:
•
Option1
Set the ADCDC1 bit in the PWR_CR register.
•
Option2
Set the corresponding ADCxDC2 bit in the SYSCFG_PMC register.
Only one option can be set at a time.
For more details on option 1 and option2 mechanisms, refer to AN4073.
2.1.8
Over-drive and Under-drive modes unavailability
Description
The Over-drive and Under-drive modes are not available on revision A devices.
Workaround
None.
2.1.9
Operating voltage extension down to 1.7 V
in the whole temperature range
Description
Revision "A" and "Y" devices powered from a 1.7 V supply voltage, operate only in the 0 to
70 °C temperature range.
Starting from revision 1, the operating voltage down to 1.7 V is extended at full temperature
ranges:
•
-40 °C to 105 °C (suffix 7)
•
-40 °C to 85 °C (for suffix 6).
Workaround
None.
DocID023833 Rev 9
13/34
33
STM32F42xx and STM32F43xx silicon limitations
2.1.10
STM32F42xx and STM32F43xx
PA12 GPIO limitation
Description
When PA12 is used as GPIO or alternate function in input or output mode, the data read
from Flash memory can be corrupted. This behavior is observed only when the following
conditions are met:
•
The device operates from a 2.7 to 3.6 V VDD power supply whatever the temperature
range
•
Flash memory Bank2 is used or the dual bank feature is enabled.
Impacted products
•
STM32F42xxI and STM32F43xxI part numbers
•
STM32F42xxG and STM32F43xxG part numbers only when dual bank feature is
enabled.
Not impacted products
•
STM32F42xxG and STM32F43xxG part numbers when dual bank feature is disabled
•
STM32F42xxE and STM32F43xxE part numbers.
Workaround
PA12 must be left unconnected on the PCB (configured as push-pull and held Low). You
can use all the other GPIOs and all alternate functions except for the ones mapped on
PA12. Use the OTG_HS peripheral in full-speed mode instead of the OTG_FS peripheral.
This limitation is fixed in silicon revision 3.
2.2
IWDG peripheral limitation
2.2.1
RVU and PVU flags are not reset in STOP mode
Description
The RVU and PVU flags of the IWDG_SR register are set by hardware after a write access
to the IWDG_RLR and the IWDG_PR registers, respectively. If the Stop mode is entered
immediately after the write access, the RVU and PVU flags are not reset by hardware.
Before performing a second write operation to the IWDG_RLR or the IWDG_PR register,
the application software must wait for the RVU or PVU flag to be reset. However, since the
RVU/PVU bit is not reset after exiting the Stop mode, the software goes into an infinite loop
and the independent watchdog (IWDG) generates a reset after the programmed timeout
period.
Workaround
Wait until the RVU or PVU flag of the IWDG_SR register is reset before entering the Stop
mode.
14/34
DocID023833 Rev 9
STM32F42xx and STM32F43xx
STM32F42xx and STM32F43xx silicon limitations
2.3
I2C peripheral limitations
2.3.1
SMBus standard not fully supported
Description
The I2C peripheral is not fully compliant with the SMBus v2.0 standard since It does not
support the capability to NACK an invalid byte/command.
Workarounds
A higher-level mechanism should be used to verify that a write operation is being performed
correctly at the target device, such as:
2.3.2
1.
Using the SMBAL pin if supported by the host
2.
the alert response address (ARA) protocol
3.
the Host notify protocol
Start cannot be generated after a misplaced Stop
Description
If a master generates a misplaced Stop on the bus (bus error), the peripheral cannot
generate a Start anymore.
Workaround
In the I²C standard, it is allowed to send a Stop only at the end of the full byte (8 bits +
acknowledge), so this scenario is not allowed. Other derived protocols like CBUS allow it,
but they are not supported by the I²C peripheral.
A software workaround consists in asserting the software reset using the SWRST bit in the
I2C_CR1 control register.
2.3.3
Mismatch on the “Setup time for a repeated Start condition” timing
parameter
Description
In case of a repeated Start, the “Setup time for a repeated Start condition” (named Tsu;sta in
the I²C specification) can be slightly violated when the I²C operates in Master Standard
mode at a frequency between 88 kHz and 100 kHz.
The limitation can occur only in the following configuration:
•
in Master mode
•
in Standard mode at a frequency between 88 kHz and 100 kHz (no limitation in Fastmode)
•
SCL rise time:
–
If the slave does not stretch the clock and the SCL rise time is more than 300 ns (if
the SCL rise time is less than 300 ns, the limitation cannot occur)
–
If the slave stretches the clock
The setup time can be violated independently of the APB peripheral frequency.
DocID023833 Rev 9
15/34
33
STM32F42xx and STM32F43xx silicon limitations
STM32F42xx and STM32F43xx
Workaround
Reduce the frequency down to 88 kHz or use the I²C Fast-mode, if supported by the slave.
2.3.4
Data valid time (tVD;DAT) violated without the OVR flag being set
Description
The data valid time (tVD;DAT, tVD;ACK) described by the I²C standard can be violated (as well
as the maximum data hold time of the current data (tHD;DAT)) under the conditions described
below. This violation cannot be detected because the OVR flag is not set (no transmit buffer
underrun is detected).
This limitation can occur only under the following conditions:
•
in Slave transmit mode
•
with clock stretching disabled (NOSTRETCH=1)
•
if the software is late to write the DR data register, but not late enough to set the OVR
flag (the data register is written before)
Workaround
If the master device allows it, use the clock stretching mechanism by programming the bit
NOSTRETCH=0 in the I2C_CR1 register.
If the master device does not allow it, ensure that the software is fast enough when polling
the TXE or ADDR flag to immediately write to the DR data register. For instance, use an
interrupt on the TXE or ADDR flag and boost its priority to the higher level.
2.3.5
Both SDA and SCL maximum rise time (tr) violated when VDD_I2C bus
higher than ((VDD+0.3) / 0.7) V
Description
When an external legacy I2C bus voltage (VDD_I2C) is set to 5 V while the MCU is powered
from VDD, the internal 5-Volt tolerant circuitry is activated as soon the input voltage (VIN)
reaches the VDD + diode threshold level. An additional internal large capacitance then
prevents the external pull-up resistor (RP) from rising the SDA and SCL signals within the
maximum timing (tr) which is 300 ns in fast mode and 1000 ns in Standard mode.
The rise time (tr) is measured from VIL and VIH with levels set at 0.3VDD_I2C and
0.7VDD_I2C.
Workaround
The external VDD_I2C bus voltage should be limited to a maximum value of
((VDD+0.3) / 0.7) V. As a result, when the MCU is powered from VDD=3.3 V, VDD_I2C
should not exceed 5.14 V to be compliant with I2C specifications.
16/34
DocID023833 Rev 9
STM32F42xx and STM32F43xx
STM32F42xx and STM32F43xx silicon limitations
2.4
I2S peripheral limitation
2.4.1
In I2S slave mode, WS level must be set by the external master
when enabling the I2S
Description
In slave mode, the WS signal level is used only to start the communication. If the I2S (in
slave mode) is enabled while the master is already sending the clock and the WS signal
level is low (for I2S protocol) or is high (for the LSB or MSB-justified mode), the slave starts
communicating data immediately. In this case, the master and slave will be desynchronized
throughout the whole communication.
Workaround
The I2S peripheral must be enabled when the external master sets the WS line at:
•
High level when the I2S protocol is selected.
•
Low level when the LSB or MSB-justified mode is selected.
2.5
USART peripheral limitations
2.5.1
Idle frame is not detected if receiver clock speed is deviated
Description
If the USART receives an idle frame followed by a character, and the clock of the transmitter
device is faster than the USART receiver clock, the USART receive signal falls too early
when receiving the character start bit, with the result that the idle frame is not detected
(IDLE flag is not set).
Workaround
None.
2.5.2
In full duplex mode, the Parity Error (PE) flag can be cleared by
writing to the data register
Description
In full duplex mode, when the Parity Error flag is set by the receiver at the end of a
reception, it may be cleared while transmitting by reading the USART_SR register to check
the TXE or TC flags and writing data to the data register.
Consequently, the software receiver can read the PE flag as '0' even if a parity error
occurred.
Workaround
The Parity Error flag should be checked after the end of reception and before transmission.
DocID023833 Rev 9
17/34
33
STM32F42xx and STM32F43xx silicon limitations
2.5.3
STM32F42xx and STM32F43xx
Parity Error (PE) flag is not set when receiving in Mute mode
using address mark detection
Description
The USART receiver is in Mute mode and is configured to exit the Mute mode using the
address mark detection. When the USART receiver recognizes a valid address with a parity
error, it exits the Mute mode without setting the Parity Error flag.
Workaround
None.
2.5.4
Break frame is transmitted regardless of nCTS input line status
Description
When CTS hardware flow control is enabled (CTSE = 1) and the Send Break bit (SBK) is
set, the transmitter sends a break frame at the end of the current transmission regardless of
nCTS input line status.
Consequently, if an external receiver device is not ready to accept a frame, the transmitted
break frame is lost.
Workaround
None.
2.5.5
nRTS signal abnormally driven low after a protocol violation
Description
When RTS hardware flow control is enabled, the nRTS signal goes high when data is
received. If this data was not read and new data is sent to the USART (protocol violation),
the nRTS signal goes back to low level at the end of this new data.
Consequently, the sender gets the wrong information that the USART is ready to receive
further data.
On USART side, an overrun is detected, which indicates that data has been lost.
Workaround
Workarounds are required only if the other USART device violates the communication
protocol, which is not the case in most applications.
Two workarounds can be used:
18/34
•
After data reception and before reading the data in the data register, the software takes
over the control of the nRTS signal as a GPIO and holds it high as long as needed. If
the USART device is not ready, the software holds the nRTS pin high, and releases it
when the device is ready to receive new data.
•
The time required by the software to read the received data must always be lower than
the duration of the second data reception. For example, this can be ensured by treating
all the receptions by DMA mode.
DocID023833 Rev 9
STM32F42xx and STM32F43xx
STM32F42xx and STM32F43xx silicon limitations
2.6
bxCAN limitation
2.6.1
bxCAN time triggered communication mode not supported
Description
The time triggered communication mode described in the reference manual is not
supported. As a result timestamp values are not available. TTCM bit must be kept cleared in
the CAN_MCR register (time triggered communication mode disabled).
Workaround
None
2.7
OTG_FS peripheral limitations
2.7.1
Data in RxFIFO is overwritten when all channels are disabled
simultaneously
Description
If the available RxFIFO is just large enough to host 1 packet + its data status, and is
currently occupied by the last received data + its status and, at the same time, the
application requests that more IN channels be disabled, the OTG_FS peripheral does not
first check for available space before inserting the disabled status of the IN channels. It just
inserts them by overwriting the existing data payload.
Workaround
Use one of the following recommendations:
2.7.2
1.
Configure the RxFIFO to host a minimum of 2 × MPSIZ + 2 × data status entries.
2.
The application has to check the RXFLVL bit (RxFIFO non-empty) in the
OTG_FS_GINTSTS register before disabling each IN channel. If this bit is not set, then
the application can disable an IN channel at a time. Each time the application disables
an IN channel, however, it first has to check that the RXFLVL bit = 0 condition is true.
OTG host blocks the receive channel when receiving IN packets and no
TxFIFO is configured
Description
When receiving data, the OTG_FS core erroneously checks for available TxFIFO space
when it should only check for RxFIFO space. If the OTG_FS core cannot see any space
allocated for data transmission, it blocks the reception channel and no data is received.
Workaround
Set at least one TxFIFO equal to the maximum packet size. In this way, the host application,
which intends to supports only IN traffic, also has to allocate some space for the TxFIFO.
Since a USB host is expected to support any kind of connected endpoint, it is good practice
to always configure enough TxFIFO space for OUT endpoints.
DocID023833 Rev 9
19/34
33
STM32F42xx and STM32F43xx silicon limitations
2.7.3
STM32F42xx and STM32F43xx
Host channel-halted interrupt not generated when the channel is
disabled
Description
When the application enables, then immediately disables the host channel before the
OTG_FS host has had time to begin the transfer sequence, the OTG_FS core, as a host,
does not generate a channel-halted interrupt. The OTG_FS core continues to operate
normally.
Workaround
Do not disable the host channel immediately after enabling it.
2.7.4
Error in software-read OTG_FS_DCFG register values
Description
When the application writes to the DAD and PFIVL bitfields in the OTG_FS_DCFG register,
and then reads the newly written bitfield values, the read values may not be correct.
The values written by the application, however, are correctly retained by the core, and the
normal operation of the device is not affected.
Workaround
Do not read from the OTG_FS_DCFG register’s DAD and PFIVL bitfields just after
programming them.
2.8
Ethernet peripheral limitations
2.8.1
Incorrect layer 3 (L3) checksum is inserted in transmitted IPv6 packets
without TCP, UDP or ICMP payloads
Description
The application provides the per-frame control to instruct the MAC to insert the L3
checksums for TCP, UDP and ICMP packets. When automatic checksum insertion is
enabled and the input packet is an IPv6 packet without the TCP, UDP or ICMP payload, then
the MAC may incorrectly insert a checksum into the packet. For IPv6 packets without a TCP,
UDP or ICMP payload, the MAC core considers the next header (NH) field as the extension
header and continues to parse the extension header. Sometimes, the payload data in such
packets matches the NH field for TCP, UDP or ICMP and, as a result, the MAC core inserts
a checksum.
Workaround
When the IPv6 packets have a TCP, UDP or ICMP payload, enable checksum insertion for
transmit frames, or bypass checksum insertion by using the CIC (checksum insertion
control) bits in TDES0 (bits 23:22).
20/34
DocID023833 Rev 9
STM32F42xx and STM32F43xx
2.8.2
STM32F42xx and STM32F43xx silicon limitations
The Ethernet MAC processes invalid extension headers in the received
IPv6 frames
Description
In IPv6 frames, there can be zero or some extension headers preceding the actual IP
payload. The Ethernet MAC processes the following extension headers defined in the IPv6
protocol: Hop-by-Hop Options header, Routing header and Destination Options header.
All extension headers, except the Hop-by-Hop extension header, can be present multiple
times and in any order before the actual IP payload. The Hop-by-Hop extension header, if
present, has to come immediately after the IPv6’s main header.
The Ethernet MAC processes all (valid or invalid) extension headers including the Hop-byHop extension headers that are present after the first extension header. For this reason, the
GMAC core will accept IPv6 frames with invalid Hop-by-Hop extension headers. As a
consequence, it will accept any IP payload as valid IPv6 frames with TCP, UDP or ICMP
payload, and then incorrectly update the Receive status of the corresponding frame.
Workaround
None.
2.8.3
MAC stuck in the Idle state on receiving the TxFIFO flush command
exactly 1 clock cycle after a transmission completes
Description
When the software issues a TxFIFO flush command, the transfer of frame data stops (even
in the middle of a frame transfer). The TxFIFO read controller goes into the Idle state
(TFRS=00 in ETH_MACDBGR) and then resumes its normal operation.
However, if the TxFIFO read controller receives the TxFIFO flush command exactly one
clock cycle after receiving the status from the MAC, the controller remains stuck in the Idle
state and stops transmitting frames from the TxFIFO. The system can recover from this
state only with a reset (e.g. a soft reset).
Workaround
Do not use the TxFIFO flush feature.
If TXFIFO flush is really needed, wait until the TxFIFO is empty prior to using the TxFIFO
flush command.
2.8.4
Transmit frame data corruption
Frame data corrupted when the TxFIFO is repeatedly transitioning from non-empty to empty
and then back to non-empty.
Description
Frame data may get corrupted when the TxFIFO is repeatedly transitioning from non-empty
to empty for a very short period, and then from empty to non-empty, without causing an
underflow.
DocID023833 Rev 9
21/34
33
STM32F42xx and STM32F43xx silicon limitations
STM32F42xx and STM32F43xx
This transitioning from non-empty to empty and back to non-empty happens when the rate
at which the data is being written to the TxFIFO is almost equal to or a little less than the
rate at which the data is being read.
This corruption cannot be detected by the receiver when the CRC is inserted by the MAC,
as the corrupted data is used for the CRC computation.
Workaround
Use the Store-and-Forward mode: TSF=1 (bit 21 in ETH_DMAOMR). In this mode, the data
is transmitted only when the whole packet is available in the TxFIFO.
2.8.5
Successive write operations to the same register might not be fully
taken into account
Description
A write to a register might not be fully taken into account if a previous write to the same
register is performed within a time period of four TX_CLK/RX_CLK clock cycles. When this
error occurs, reading the register returns the most recently written value, but the Ethernet
MAC continues to operate as if the latest write operation never occurred.
See Table 5: Impacted registers and bits for the registers and bits impacted by this limitation.
Table 5. Impacted registers and bits
Register name
Bit number
Bit name
7
EDFE
26
DTCEFD
25
RSF
20
FTF
7
FEF
6
FUGF
4:3
RTC
DMA registers
ETH_DMABMR
ETH_DMAOMR
GMAC registers
22/34
DocID023833 Rev 9
STM32F42xx and STM32F43xx
STM32F42xx and STM32F43xx silicon limitations
Table 5. Impacted registers and bits (continued)
Register name
ETH_MACCR
Bit number
Bit name
25
CSTF
23
WD
22
JD
19:17
IFG
16
CSD
14
FES
13
ROD
12
LM
11
DM
10
IPCO
9
RD
7
APCS
6:5
BL
4
DC
3
TE
2
RE
ETH_MACFFR
MAC frame filter register
ETH_MACHTHR
31:0
Hash Table High Register
ETH_MACHTLR
31:0
Hash Table Low Register
31:16
PT
7
ZQPD
5:4
PLT
3
UPFD
2
RFCE
1
TFCE
0
FCB/BPA
16
VLANTC
15:0
VLANTI
ETH_MACFCR
ETH_MACVLANTR
ETH_MACRWUFFR
ETH_MACPMTCSR
all remote wakeup registers
31
WFFRPR
9
GU
2
WFE
1
MPE
0
PD
ETH_MACA0HR
MAC address 0 high register
DocID023833 Rev 9
23/34
33
STM32F42xx and STM32F43xx silicon limitations
STM32F42xx and STM32F43xx
Table 5. Impacted registers and bits (continued)
Register name
Bit number
Bit name
ETH_MACA0LR
MAC address 0 low register
ETH_MACA1HR
MAC address 1 high register
ETH_MACA1LR
MAC address 1 low register
ETH_MACA2HR
MAC address 2 high register
ETH_MACA2LR
MAC address 2 low register
ETH_MACA3HR
MAC address 3 high register
ETH_MACA3LR
MAC address 3 low register
IEEE 1588 time stamp registers
ETH_PTPTSCR
18
TSPFFMAE
17:16
TSCNT
15
TSSMRME
14
TSSEME
13
TSSIPV4FE
12
TSSIPV6FE
11
TSSPTPOEFE
10
TSPTPPSV2E
9
TSSSR
8
TSSARFE
5
TSARU
3
TSSTU
2
TSSTI
1
TSFCU
0
TSE
Workaround
Two workarounds could be applicable:
24/34
•
Ensure a delay of four TX_CLK/RX_CLK clock cycles between the successive write
operations to the same register.
•
Make several successive write operations without delay, then read the register when all
the operations are complete, and finally reprogram it after a delay of four
TX_CLK/RX_CLK clock cycles.
DocID023833 Rev 9
STM32F42xx and STM32F43xx
STM32F42xx and STM32F43xx silicon limitations
2.9
FMC peripheral limitation
2.9.1
Dummy read cycles inserted when reading synchronous memories
Description
When performing a burst read access to a synchronous memory, two dummy read accesses
are performed at the end of the burst cycle whatever the type of AHB burst access.
However, the extra data values which are read are not used by the FMC and there is no
functional failure.
Workaround
None.
2.9.2
FMC synchronous mode and NWAIT signal disabled
Description
When the FMC synchronous mode operates with the NWAIT signal disabled, if the polarity
(WAITPOL in the FMC_BCRx register) of the NWAIT signal is identical to that of the NWAIT
input signal level, the system hangs and no fault is generated.
Workaround
PD6 (NWAIT signal) must not be connected to AF12 and the NWAIT polarity must be
configured to active high (set WAITPOL bit to 1 in FMC_BCRx register).
2.9.3
Read access to a non-initialized FMC_SDRAM bank
Description
When a read access is performed to an SDRAM bank while the SDRAM controller is not yet
initialized, the system hangs and no fault is generated.
Workaround
Read access to an SDRAM bank must be performed only when the SDRAM controller
initialization is complete.
2.9.4
Corruption of data read from the FMC
Description
When the FMC is used as stack, heap or variable data, an interrupt occurring during a CPU
read access to the FMC may results in read data corruption or hard fault exception. This
problem does not occur when read accesses are performed by another master or when
FMC accesses are done when the interrupts are disabled.
DocID023833 Rev 9
25/34
33
STM32F42xx and STM32F43xx silicon limitations
STM32F42xx and STM32F43xx
Workaround
Two workarounds can be applied:
•
Do not use the FMC as stack or heap, and make sure CPU read accesses to the FMC
are performed while interrupts are disabled
•
Use only DMAs to perform read accesses to the FMC.
This limitation is present only in revision “A” devices. It is fixed in revision “Y”, “1” and “3”.
2.9.5
Interruption of CPU read burst access to an end of SDRAM row
Description
If an interrupt occurs during an CPU AHB burst read access to an end of SDRAM row, it
may result in wrong data read from the next row if all the conditions below are met:
•
The SDRAM data bus is 16-bit or 8-bit wide. 32-bit SDRAM mode is not affected.
•
RBURST bit is reset in the FMC_SDCR1 register (read FIFO disabled).
•
An interrupt occurs while CPU is performing an AHB incrementing bursts read access
of unspecified length (using LDM = Load Multiple instruction).
•
The address of the burst operation includes the end of an SDRAM row.
Workaround
Enable the read FIFO by setting the RBURST bit in the FMC_SDCR1 register.
2.9.6
FMC NOR/PSRAM controller: asynchronous read access on bank 2 to 4
returns wrong data when bank 1 is in synchronous mode
(BURSTEN bit is set)
Description
If an interrupt occurs during a CPU AHB read access to one NOR/PSRAM bank (bank2 to 4)
which is enabled in asynchronous mode, while bank 1 of the NOR/PSRAM controller is
configured in synchronous read mode (BURSTEN bit set to ‘1’), then the FMC NOR/PSRAM
controller returns wrong data.
This limitation does not occur when using the DMA or when only bank1 is used in
synchronous mode.
Workaround
If multiple banks are enabled in mixed asynchronous and synchronous modes, use any
NOR/PSRAM bank for synchronous read accesses, except for bank 1. As a consequence
the continuous clock feature is not available in asynchronous mode.
This limitation is fixed in revision “1” and “3”.
26/34
DocID023833 Rev 9
STM32F42xx and STM32F43xx
2.9.7
STM32F42xx and STM32F43xx silicon limitations
FMC dynamic and static bank switching
Description
The dynamic and static banks cannot be accessed concurrently.
Workaround
Do not use dynamic and static banks at the same time. The SDRAM device must be in selfrefresh before switching to the static memory mapped on the NOR/PSRAM or NAND/PCCard controller. Before switching from static memory to SDRAM, issue a Normal command
to wake-up the device from self-refresh mode.
This limitation is fixed in silicon revision “3”.
2.9.8
NAND/PCCard transaction and Wait timing
Description
If the WAIT timing in the corresponding space (common/attribute) is configured to 0xFF, the
NAND/PCCard transaction stalls the system and no fault is generated. This issue occurs
whatever the value of PWAITEN bit in the FMC_PCRx register.
Workaround
Configure the WAIT timing in the (common/attribute) to any value except 0xFF.
2.9.9
Data corruption during burst read from FMC synchronous memory
Description
A burst read from static memory can be corrupted if all the following conditions are met:
•
One FMC bank is configured in synchronous mode with WAITEN bit enabled while
another FMC bank is used with WAITEN bit disabled
•
A read burst transaction is ongoing from static synchronous memory with wait feature
enabled
•
The synchronous memory asserts the wait signal during the ongoing burst read
•
The read burst transaction is followed by an access to an FMC dynamic or static banks
which the WAITEN bit is disabled in the FMC_BCRx register.
Workaround
1.
Set the WAITEN bit on all FMC static banks even if it is not used by the memory.
2.
Set the same WAIT polarity on all static banks
3.
Enable the internal pull-up on PD6 in order to set to ready the FMC_NWAIT input when
the synchronous memory is de-selected and the other FMC bank without wait feature is
selected.
DocID023833 Rev 9
27/34
33
STM32F42xx and STM32F43xx silicon limitations
2.9.10
STM32F42xx and STM32F43xx
Missed burst write transaction on multiplexed PSRAM
Description
If an interrupt occurs during an CPU AHB read access from an FMC bank with the write
burst feature disabled (CBURSTRW bit is reset in the FMC_BCRx register), the following
burst write access to another FMC bank with the write feature enabled is missed.
Workaround
Set the CBURSTRW bit even for the FMC bank that is not targeted by the write burst
access. The write burst transaction has no effect on the asynchronous read protocol.
However, to perform write accesses to asynchronous FMC bank while the CBURSTRW bit
is set, the AHB size must be equal to memory size.
2.9.11
FMC NOR/PSRAM controller write protocol violation
Description
When an interrupted asynchronous or synchronous CPU read access to any NOR/PSRAM
FMC bank is followed by an asynchronous write access to the same bank or to any other
NOR/PSRAM FMC bank, this causes a write protocol violation. There is no functional issue
but the FMC NOR/PSRAM controller write protocol is violated since the FMC_NWE signal is
de-asserted at the same time as the Chip select (FMC_NEx).
Workaround
None.
2.9.12
FMC NOR/PSRAM controller bank switch with different
BUSTURN durations
Description
The system hangs when the FMC NOR/PSRAM memory controller switches between two
banks and the following conditions are met:
1.
One NOR/PSRAM bank is configured in synchronous mode and the BUSTURN bits in
FMC_BTRx/FMC_BTWx register are set to a nonzero value.
2.
Another NOR/PSRAM bank is configured in asynchronous multiplexed mode and
BUSTURN bits are set to 0.
3.
FMC clock divide ratio (CLKDIV) is higher or equal to 4 HCLK periods.
4.
A single read transaction from the bank operating in synchronous mode is followed by
any transaction in another bank operating in asynchronous multiplexed mode.
Workaround
If several NOR/PSRAM banks are used, the BUSTURN duration configured for each bank
must be set to a nonzero value.
28/34
DocID023833 Rev 9
STM32F42xx and STM32F43xx
STM32F42xx and STM32F43xx silicon limitations
2.10
SDIO peripheral limitations
2.10.1
SDIO HW flow control
Description
When enabling the HW flow control by setting bit 14 of the SDIO_CLKCR register to ‘1’,
glitches can occur on the SDIOCLK output clock resulting in wrong data to be written into
the SD/MMC card or into the SDIO device. As a consequence, a CRC error will be reported
to the SD/SDIO MMC host interface (DCRCFAIL bit set to ‘1’ in SDIO_STA register).
Workaround
None.
Note:
Do not use the HW flow control. Overrun errors (Rx mode) and FIFO underrun (Tx mode)
should be managed by the application software.
2.10.2
Wrong CCRCFAIL status after a response without CRC is received
Description
The CRC is calculated even if the response to a command does not contain any CRC field.
As a consequence, after the SDIO command IO_SEND_OP_COND (CMD5) is sent, the
CCRCFAIL bit of the SDIO_STA register is set.
Workaround
The CCRCFAIL bit in the SDIO_STA register shall be ignored by the software. CCRCFAIL
must be cleared by setting CCRCFAILC bit of the SDIO_ICR register after reception of the
response to the CMD5 command.
2.10.3
Data corruption in SDIO clock dephasing (NEGEDGE) mode
Description
When NEGEDGE bit is set to ‘1’, it may lead to invalid data and command response read.
Workaround
None. A configuration with the NEGEDGE bit equal to ‘1’ should not be used.
2.10.4
CE-ATA multiple write command and card busy signal management
Description
The CE-ATA card may inform the host that it is busy by driving the SDIO_D0 line low, two
cycles after the transfer of a write command (RW_MULTIPLE_REGISTER or
RW_MULTIPLE_BLOCK). When the card is in a busy state, the host must not send any
data until the BUSY signal is de-asserted (SDIO_D0 released by the card).
This condition is not respected if the data state machine leaves the IDLE state (Write
operation programmed and started, DTEN = 1, DTDIR = 0 in SDIO_DCTRL register and
TXFIFOE = 0 in SDIO_STA register).
As a consequence, the write transfer fails and the data lines are corrupted.
DocID023833 Rev 9
29/34
33
STM32F42xx and STM32F43xx silicon limitations
STM32F42xx and STM32F43xx
Workaround
After sending the write command (RW_MULTIPLE_REGISTER or
RW_MULTIPLE_BLOCK), the application must check that the card is not busy by polling the
BSY bit of the ATA status register using the FAST_IO (CMD39) command before enabling
the data state machine.
2.10.5
No underrun detection with wrong data transmission
Description
In case there is an ongoing data transfer from the SDIO host to the SD card and the
hardware flow control is disabled (bit 14 of the SDIO_CLKCR is not set), if an underrun
condition occurs, the controller may transmit a corrupted data block (with wrong data word)
without detecting the underrun condition when the clock frequencies have the following
relationship:
[3 x period(PCLK2) + 3 x period(SDIOCLK)] >= (32 / (BusWidth)) x period(SDIO_CK)
Workaround
Avoid the above-mentioned clock frequency relationship, by:
•
Incrementing the APB frequency
•
or decreasing the transfer bandwidth
•
or reducing SDIO_CK frequency
2.11
ADC peripheral limitations
2.11.1
ADC sequencer modification during conversion
Description
If an ADC conversion is started by software (writing the SWSTART bit), and if the
ADC_SQRx or ADC_JSQRx registers are modified during the conversion, the current
conversion is reset and the ADC does not restart a new conversion sequence automatically.
If an ADC conversion is started by hardware trigger, this limitation does not apply. The ADC
restarts a new conversion sequence automatically.
Workaround
When an ADC conversion sequence is started by software, a new conversion sequence can
be restarted only by setting the SWSTART bit in the ADC_CR2 register.
30/34
DocID023833 Rev 9
STM32F42xx and STM32F43xx
STM32F42xx and STM32F43xx silicon limitations
2.12
DAC peripheral limitations
2.12.1
DMA underrun flag management
Description
If the DMA is not fast enough to input the next digital data to the DAC, as a consequence,
the same digital data is converted twice. In these conditions, the DMAUDR flag is set, which
usually leads to disable the DMA data transfers. This is not the case: the DMA is not
disabled by DMAUDR=1, and it keeps servicing the DAC.
Workaround
To disable the DAC DMA stream, reset the EN bit (corresponding to the DAC DMA stream)
in the DMA_SxCR register.
2.12.2
DMA request not automatically cleared by DMAEN=0
Description
if the application wants to stop the current DMA-to-DAC transfer, the DMA request is not
automatically cleared by DMAEN=0, or by DACEN=0.
If the application stops the DAC operation while the DMA request is high, the DMA request
will be pending while the DAC is reinitialized and restarted; with the risk that a spurious
unwanted DMA request is serviced as soon as the DAC is re-enabled.
Workaround
To stop the current DMA-to-DAC transfer and restart, the following sequence should be
applied:
1.
Check if DMAUDR is set.
2.
Clear the DAC/DMAEN bit.
3.
Clear the EN bit of the DAC DMA/Stream
4.
Reconfigure by software the DAC, DMA, triggers etc.
5.
Restart the application.
DocID023833 Rev 9
31/34
33
Revision history
3
STM32F42xx and STM32F43xx
Revision history
Table 6. Document revision history
32/34
Date
Revision
Changes
11-Feb-2013
1
Initial release.
25-Feb-2013
2
Document converted to new template.
Added Section 2.9.4: Corruption of data read from the FMC
26-Apr-2013
3
Added Silicon revision Y.
Removed the reference to ‘Cortex-M4F’ in the whole document.
Updated Section 2.9.1: Dummy read cycles inserted when reading
synchronous memories.
Added Section 2.1.3: Wakeup sequence from Standby mode when
using more than one wakeup source, Section 2.8.5: Successive write
operations to the same register might not be fully taken into account
and Section 2.8.3: FSMC NOR Flash/PSRAM controller
asynchronous access on bank 2 to 4 when bank 1 is in synchronous
mode (CBURSTRW bit is set).
Removed limitation 2.10.3 SDIO clock divider BYPASS mode may
not work properly. Updated Section 2.10.5: No underrun detection
with wrong data transmission.
xx-xxx-2014
4
Added Section 2.6.1: bxCAN time triggered communication mode
not supported.
19-Sep-2013
4
Added STM32F429xx and STM32F439xx devices.
Removed FSMC limitations.
Added Section 2.3.5: Both SDA and SCL maximum rise time (tr)
violated when VDD_I2C bus higher than ((VDD+0.3) / 0.7) V.
Updated Section 2.9.5: Interruption of CPU read burst access to an
end of SDRAM row.
Added Section 2.9.1: Dummy read cycles inserted when reading
synchronous memories, Section 2.9.2: FMC synchronous mode and
NWAIT signal disabled, Section 2.9.3: Read access to a noninitialized FMC_SDRAM bank, Section 2.9.4: Corruption of data read
from the FMC, Section 2.9.5: Interruption of CPU read burst access
to an end of SDRAM row, Section 2.9.6: FMC NOR/PSRAM
controller: asynchronous read access on bank 2 to 4 returns wrong
data when bank 1 is in synchronous mode (BURSTEN bit is set) and
Section 2.9.7: FMC dynamic and static bank switching.
Added Figure 1: TFBGA216 top package view, Figure 2: WLCSP143
top package view, and Figure 3: LQFP208 top package view.
23-Sep-2013
5
Updated workaround in Section 2.9.6: FMC NOR/PSRAM controller:
asynchronous read access on bank 2 to 4 returns wrong data when
bank 1 is in synchronous mode (BURSTEN bit is set).
DocID023833 Rev 9
STM32F42xx and STM32F43xx
Revision history
Table 6. Document revision history (continued)
Date
09-Jan-2014
05-May-2014
20-Jun-2014
03-Oct-2014
Revision
Changes
6
Added silicon revision 1.
Added STM32F429xE, STM32F427Ax, STM32F437Ax,
STM32F429Ax, and STM32F439Ax part numbers.
Removed mention of limitation fix in Section 2.1.8: Over-drive and
Under-drive modes unavailability, Section 2.9.4: Corruption of data
read from the FMC and Section 2.9.6: FMC NOR/PSRAM controller:
asynchronous read access on bank 2 to 4 returns wrong data when
bank 1 is in synchronous mode (BURSTEN bit is set).
Updated Section 2.9.7: FMC dynamic and static bank switching to
indicate the limitation will be fixed in next silicon revision.
7
Added silicon revision 3.
Added Section 1.2: VDIV or VSQRT instructions might not complete
correctly when very short ISRs are used.
Added Section 2.1.9: Operating voltage extension down to 1.7 V in
the whole temperature range, Section 2.9.8: NAND/PCCard
transaction and Wait timing, Section 2.9.9: Data corruption during
burst read from FMC synchronous memory, and Section 2.9.10:
Missed burst write transaction on multiplexed PSRAM.
Moved all device marking schematics to datasheets.
8
Added Section 2.1.10: PA12 GPIO limitation, Section 2.9.11: FMC
NOR/PSRAM controller write protocol violation and Section 2.9.12:
FMC NOR/PSRAM controller bank switch with different BUSTURN
durations.
9
Updated FMC NOR/PSRAM controller write protocol violation
limitation and FMC synchronous mode and NWAIT signal disabled
limitations in Table 4: Summary of silicon limitations.
Updated Section 2.1.10: PA12 GPIO limitation.
Added Section 2.6.1: bxCAN time triggered communication mode
not supported.
DocID023833 Rev 9
33/34
33
STM32F42xx and STM32F43xx
IMPORTANT NOTICE – PLEASE READ CAREFULLY
STMicroelectronics NV and its subsidiaries (“ST”) reserve the right to make changes, corrections, enhancements, modifications, and
improvements to ST products and/or to this document at any time without notice. Purchasers should obtain the latest relevant information on
ST products before placing orders. ST products are sold pursuant to ST’s terms and conditions of sale in place at the time of order
acknowledgement.
Purchasers are solely responsible for the choice, selection, and use of ST products and ST assumes no liability for application assistance or
the design of Purchasers’ products.
No license, express or implied, to any intellectual property right is granted by ST herein.
Resale of ST products with provisions different from the information set forth herein shall void any warranty granted by ST for such product.
ST and the ST logo are trademarks of ST. All other product or service names are the property of their respective owners.
Information in this document supersedes and replaces information previously supplied in any prior versions of this document.
© 2014 STMicroelectronics – All rights reserved
34/34
DocID023833 Rev 9