This patch modifies the OTA implementation to incremental erase flash sectors
as they are written, rather than bulk erasing them all at once when OTA begins.
This solves several problems:
- The bulk flash erase can take so long that it triggers the watchdog timer.
Disabling the watchdog is an undesirable risk.
- The bulk erase interferes with the responsiveness of the system to user input.
- Incremental erasing eliminates the need to pass the firmware image size to
esp_ota_begin to avoid erasing more flash sectors than required.
- When installing an OTA received via HTTP, the size of the firmware image is
known when the content-length header is received. Calling esp_ota_begin at that
time causes a long delay to do the bulk erase which causes the TCP transfer
speed to be measurably slowed when the transfer resumes after the erase.
Incremental erasing eliminates this TCP slowdown.
Update:
Rework so erase optimization is only applied when image_size is set to
OTA_WITH_SEQUENTIAL_WRITES #5246
Merges https://github.com/espressif/esp-idf/pull/5246
Summary of changes:
- bootloader_clock split into *_clock_init and *_clock_loader.
Only esp_clk_apb_freq is in *_clock_loader.
- bootloader_common moved out of loader; functions needed in loader
(or, referenced from bootloader_utility) were moved into
bootloader_common_loader.c.
- assert and abort moved into bootloader_panic, made part of the
loader
- rtc_clk and rtc_time made part of loader
- Previously when a model is initialized or deinitialized, in the
access layer, we need to check the model id with the ids in the
table in order to find the proper model operation function.
- Currently all the operation functions of each model will be set
during the mesh initialization. When the model is found, we can
directly use the corresponding callback for different operations.
- Currently only init/deinit operations are registered, later we
will add more operations.
Spinlocks from spinlock.h do not disable the scheduler and thus cannot safely
be directly used as a locking mechanism. A task holding the lock can get
pre-empted, and at that point the new running task will also be allowed to
take the spinlock and access whatever it was protecting.
Another issue is that the task holding a spinlock could migrate to a different
core which in turn would cause the application to fail asserts. The current
implementation assumes the core that takes the lock is also the core that
releases it.
Closes https://github.com/espressif/esp-idf/issues/5762
Commit 891eb3b0 was fixing an issue with PS and EPC1 not being
preserved after the window spill procedure. It did so by saving PS in
a2 and EPC1 in a4. However the a4 register may be a live register of
another window in the call stack, and if it is overwritten and then
spilled to the stack, then the corresponding register value will end
up being corrupted. In practice the problem would show up as an
IllegalInstruction exception, when trying to return from a function
when a0 value was 0x40020.
Fix by using a0 register instead of a4 as scratch. Also fix a comment
about xthal_save_extra_nw, as this function in fact doesn't clobber
a4 or a5 because XCHAL_NCP_NUM_ATMPS is defined as 1.
Closes https://github.com/espressif/esp-idf/issues/5758
By default adds original queries to responses in order to be resolved by some resolvers, such as lwIP mdns library. This functionality however is discouraged by the RFC6762, so it could be disabled in menuconfig if MDNS_STRICT_MODE configured
Closes https://github.com/espressif/esp-idf/issues/5521
H/W decryption of Mgmt frames was disabled for PMF and done through
S/W. If ESPNOW packets go through this path, it affects backward
compatibility since method of decrypting Mgmt packets is different in H/W.
To address PMF + ESPNOW Co-existance, CCMP decryption method is modified
for ESPNOW packets so that they can be decrypted correctly. Since Tx
of ESPNOW packets can still be done in H/W alongside PMF, no change
required in encryption method in S/W.
Co-Authored-By: Nachiket Kukade <nachiket.kukade@espressif.com>
Co-Authored-By: zhangyanjiao <zhangyanjiao@espressif.com>
Co-Authored-By: kapil.gupta <kapil.gupta@espressif.com>
Sometimes the flash size read from bootloader is not correct. This may
forbid SPI Flash driver from reading the the area larger than the size
in bootloader header.
When the new config option is enabled, the latest configured
ESPTOOLPY_FLAHSIZE in the app header will be used to override the value
read from bootloader header.
