If httpd_sess_trigger_close() gets called twice for the same socket,
the first httpd_sess_close() cb closes the correct socket, but the second
invocation closes the wrong socket which was just accepted and added to
the socket db. Checking for the lru counter will help identify this as the
counter is set only for requests actually served.
We fixed some flash bugs in bootloader, but for the users used the old
vrsion bootloader, they can not fix these bugs via OTA, the solution is
add these updates in app startup.
These updates include:
1. SPI flash gpio matrix and drive strength configuration
2. SPI flash clock configuration
3. SPI flash read dummy configuration
4. SPI flash cs timing configuration
5. Update flash id of g_rom_flashchip
Hardware AES-CBC performance changes:
Release config 11.0MB/sec -> 10.8MB/sec
Debug config 9.4MB/sec -> 9.8MB/sec
(Unrolling the loop to optimize the check improves
performance at -Og, even with the fault check.)
Made MMU pages available check in `esp_image_format.c`
This now makes it possible to map and process bootoader image as well in chunks when image doesn't fit completely into available free pages.
Signed-off-by: Vikram Dattu <vikram.dattu@espressif.com>
There are lot of prints of `Allocate TX buffer for DMA`
Changed these from `ESP_LOGI` to `ESP_LOGD`
Signed-off-by: Vikram Dattu <vikram.dattu@espressif.com>
This problem may cause HCI send command timeout.
When host call VHCI api to do btdm_wakeup_request, then controller
wakeup process will be handled in ISR and controller task context.
As host task priority is lower than controller task and ISR,
it will cause an incorrect behavior that before VHCI take the
rx_flow_on_semaphore, controller sleep again, then VHCI cannot take
the semaphore and has to wait the automatic wakeup.
On flash program operation (either erase or write), if corresponding address has
cache mapping present then cache is explicitly flushed (for both pro and app cpu)
Closes https://github.com/espressif/esp-idf/issues/2146
These changes guarantee that the SSID and Passphrase received via protocomm are NULL terminated and size limited to their standard lengths.
List of changes:
* Corrected length of passphrase field in wifi_prov_config_set_data_t structure
* Performing length checks on SSID, passphrase and bssid, when populating wifi_prov_config_set_data_t structure with received credentials
1. resend eb which are pending in g_ic.tx_buf due to channel switch.
2. add lock for mgmg/xon/bcast to protect send from user flush.
3. softap: add check if secondary offset is correct when start softap.
4. softap: in inactive timer handler, add check if child bss is removed to avoid removing it again.
5. disable A-MPDU for non-root.
6. fix reusing a freed eb.
7. fix when node becomes root, it does not record assoc value to nvs.
8. discard unnecessary forwarded group packets.
9. fix toDS state is not updated in time.
10. fix s_sta_cnx_times is not cleared when deinit causes root sends ADD annoucement.
11. root: increase beacon timeout time from 6 seconds to 15 seconds to delay initiation of reconnection.
12. add esp_mesh_get_tsf_time to return TSF time.
13. send the whole routing table multi-times.
14. clear candidate settings if not receive candidate beacon within 8 seconds.
When flash work in DIO Mode, in order to ensure the fast read mode of flash
is a fixed value, we merged the mode bits into address part, and the fast
read mode value is 0 (the default value).
open_fn() was introduced in the context of HTTPS server, as a configurable callback function that is called by the HTTP server, on every newly created socket. It is responsible of allocating resources for per session transport security.
Earlier, if open_fn were to fail, the newly created socket would be closed by the server but the corresponding entry, for the now invalid socket, will remain in the internal socket database until that invalid socket is detected due to error when calling select(). Because of this delayed closing of sockets, the HTTPS server would quickly face shortage of available sessions when a lot of SSL handshake errors are happening (this typically occurs when a browser finds that the server certificate is self signed). This changes in this MR fix this issue by clearing up the socket from internal database, right after open_fn fails.
Closes https://github.com/espressif/esp-idf/issues/3479
List of changes in components/wifi_provisioning:
* Manager version is now v1.1
* .proto files and protocomm handler added for sending Wi-Fi scan command and receiving scan results
* Implemented handlers for wifi_scan protocomm endpoint
* Update manager context data structure to hold scan state and results
* scheme_softap now runs Wi-Fi in APSTA mode
* Wi-Fi is started in AP mode when provisioning is started. This is necessary for scan list to work
* Docs updates with information about new wifi_scan endpoint
List of changes in tools/esp_prov:
* Added functions for sending and receiving protobuf messages compatible with wifi_scan protocomm endpoint
* Added feature to display/refresh scan results and accept user selection at runtime
* New functions:
* get_version() : only returns the protocol version string
* has_capability() : check is a capability is present according to proto-ver response
* wifi_scan feature is provided only if the `wifi_scan` capability is present
Other changes:
* Replace recursive mutex with plain mutex
* assert on return value of mutex give / take calls
* replace all calls with macros ACQUIRE_LOCK and RELEASE_LOCK
* some checks added in scanning related private APIs
* free and nullify scanning context and state if service is stopped while ongoing scan
port*_CRITICAL_SAFE API calls port*_CRITICAL or port*_CRITICAL_ISR
depending on the context (Non-ISR or ISR respectively).
FREERTOS_CHECK_PORT_CRITICAL_COMPLIANCE Kconfig option added
Signed-off-by: Sachin Parekh <sachin.parekh@espressif.com>
xTaskIncrementTick have to unwind uxPendedTicks on CPU1 and CPU0.
Use case: If an erase operation was run on the CPU1 then it leads
to starving other tasks which waiting time. Waited tasks just skipped.
Closes: https://github.com/espressif/esp-idf/issues/1952
Closes: IDF-183
Fixed the case when the first part of log was missed
this was happened when:
* CONFIG_CONSOLE_UART_CUSTOM option is selected (UART1)
* The selected CONSOLE_UART port is used also for the console component
* in code esp_restart() or abort() functions were called.
List of changes:
* Use 128 bit characteristic UUIDs when creating GATT table entries
* Change primary service attribute value to 128 bit custom service UUID
* Use raw advertisement data to convey flags and 128 bit primary service UUID
* Use raw scan response to send device name as complete local name
* Increase maximum device name length in relation to maximum scan response length
* Set Characteristic User Description attributes for each characteristic to convey protocomm endpoint names
List of changes:
* Kconfig option HTTPD_LOG_PURGE_DATA enables logging of purged data
* Kconfig option HTTPD_PURGE_BUF_LEN sets purge buffer length
* Purged data is logged in hex
Closes https://github.com/espressif/esp-idf/issues/3359
Fix following wpa2 enterprise vulnerability issues:
1. The station can complete 4-way handshake after EAP-FAIL is received
2. The station crashes if EAP-SUCCESS is received before PMK is setup
Issue is that when users creates a custom bootloader from
$IDF_PATH/components/bootloader. Parent project build uses the copy but
bootloader subproject build uses the original still. The issue is solved
by passing the custom bootloader as extra component directory so
bootloader build knows to use the new copy (itself) in the build.
This prevents a device from being bricked in case when both secure boot & flash encryption are enabled and encryption gets interrupted during first boot. After interruption, all partitions on the device need to be reflashed (including the bootloader).
List of changes:
* Secure boot key generation and bootloader digest generation logic, implemented inside function esp_secure_boot_permanently_enable(), has been pulled out into new API esp_secure_boot_generate_digest(). The enabling of R/W protection of secure boot key on EFUSE still happens inside esp_secure_boot_permanently_enable()
* Now esp_secure_boot_permanently_enable() is called only after flash encryption process completes
* esp_secure_boot_generate_digest() is called before flash encryption process starts
Issue : Restarting the server without 30sec delay between httpd_stop() and httpd_start() causes EADDRINUSE error
Resolution : Use setsockopt() to enable SO_REUSEADDR on listener socket
Closes https://github.com/espressif/esp-idf/issues/3381
Fixed adjtime function: While using the adjtime() function,
the time correction accumulated an error
when reading the time frequently (using gettimeofday).
When using direct operations instead of `esp_http_client_perform`, we need a way to set redirection URL when we get 30x response codes. Added the function for the same.
User can now check status code and call `esp_http_client_set_redirection` function to enable redirection.
Related change in adf: https://gitlab.espressif.cn:6688/adf/esp-adf-internal/merge_requests/187
Closes https://github.com/espressif/esp-idf/issues/2712
Signed-off-by: Vikram Dattu <vikram.dattu@espressif.com>
Operation:
In `esp_http_client_set_url`, we check for if old_host is same as new_host.
Delete and open new connection if host is different.
Issue:
We just pointed `client->connection_info.host` to `old_host` and reassigned it.
This made old_host and new_host always point to same location and hence, using old_host with new request.
Fix:
Made a separate copy for old_host using strdup.
Closes https://github.com/espressif/esp-idf/issues/2631
Signed-off-by: Vikram Dattu <vikram.dattu@espressif.com>
1. Softap support multicast and broadcast saving and flushing when the associated stations enable IEEE80211 legacy power save(which is called modem sleep in ESP32).
2. Improve the frame saving and flushing mechanism of softap power save.
Earlier eraseItem function in Storage class would do lazy cleanup of
multi-page blobs if called using type "ANY" instead of "BLOB". It used to
just delete BLOB data and index would remain as is. Any subsequent read
would delete index entry as well. This however would return a valid
length without error if nvs_get_blob API was just used for finding
length and not reading the complete blob. This change fixes this issue.
Closes https://github.com/espressif/esp-idf/issues/3255
1. remove redundant SPI clock settings, use rom functions to set clock.
2. remove redundant SPI cs setup and hold settings.
3. for old 32Mbit psram, cs hold time must only be 0.5T due to the special driving mode.(cs_setup = 0; cs_hold = 0)
4. for new 64Mbit psram, cs hold time is recommended to be 2.5T. (cs_setup = 1, cs_setup_time = 0;cs_hold = 1, cs_hold_time = 1)
1. Pass the correct conn_id to protocomm_req_handle
In transport_simple_ble_write(), passing param->exec_write.conn_id would
be invalid. Instead param->write.conn_id should be passed
Similar change in transport_simple_ble_exec_write() to use
param->exec_write.conn_id
2. simple_ble_start() assumes that the mode is BLE only and enables
Bluetooth controller accordingly. For, cases having BT + BLE like Alexa
(Provisioning over BLE + Audio over classic BT), this assumption should
be removed.
Linker script generator produces build/esp32/esp32.common.ld from
components/esp32/ld/esp32.common.ld.in
This works fine until IDF is downgraded to V3.1 which uses components/esp32/ld/esp32.common.ld and
doesn't track build/esp32/esp32.common.ld at all.
At this point, the linker runs in the build/esp32 directory and "-T esp32.common.ld" picks up the
linker script generated .ld file, which causes mis-builds.
As reported on forums: https://esp32.com/viewtopic.php?f=13&t=9684&p=40105
This fix prevents HTTP server from accepting new connections when the total count of connected
sockets has reached the max_open_sockets limit set during configuration. The pending connections
are kept in backlog until atleast one of the connected sockets is closed. The maximum number of
connection requests that can kept in backlog is specified as backlog_conn configuration option.
Note that this modification has no effect when LRU purge is enabled.
Also added sanity check on setting for max_open_sockets during configuration.
Solution suggested by jimparis https://github.com/espressif/esp-idf/issues/3183#issue-421234265
Closes https://github.com/espressif/esp-idf/issues/3183
Fix following STA WFA certification issues:
1. STA HT2040 coexist
2. Disallow WEP/TKIP with HT rates
3. WPA2 improvement
4. Minor refactor for WiFi internal global variable auth_type
List of changes:
* When parsing requests, count termination from LF characters only
* Correct memcpy() length parameter in httpd_unrecv() (pointed out by jimparis in GitHub issue thread)
* Use ssize_t to store results of length subtractions during parsing
* Modify some comments to reduce ambiguity
Closes https://github.com/espressif/esp-idf/issues/3182
Issue: MDNS server initially sends probing packets to resolve naming confilicts with already registered service instances. In case of a conflict, instance name is altered and probing restarts. Original instance however wasnnot removed from the structure and upon service removal only one entry was removed and a dangling service might have been kept in the structure to bring about a crash.
Resolution: Keep only one instance of a service in the probing structure.
Closes IDF-498
possible race condition: user task runs mdns_init, which enables pcbs while mdns-task already created could execute enable/disable of the same pcbs if an appropriate system event received
mnds_free() initiates stop and delete timer tasks, which after locking the mutex could lead to a dead lock in case timer task executed before deleting the task, as it would wait indefinitelly for unlocking the mutex. This condition is fixed by calling _mdns_stop_timer without locking the mutex, because there's no need to protect any data when stopping and deleting the timer task
Closes https://github.com/espressif/esp-idf/issues/1696
Issue: mdns_service_add API allocates and queues an action to be processed in mdns task context; when allocation or queueing fails, allocated structure needs to be freed. Function _mdns_free_service did not only fee all the structures, but also updates packet data.
Resolution: Moved removal of packet data outside of _mdns_free_service function.
packets scheduled to transmit are pushed to action queue and removed from tx_queue_head structure, which is searched for all remaining services and while service is removed, then service questions/asnwers are also removed from this structure. This update fixes possible crash when packet is pushed to action queue, and when service is removed, its answers are removed from tx_queue_head, but not from action queue. this could lead to a crash when the packet is poped from action queue containing questions/answers to already removed (freed) service
Closes IDF-504
mdns_search_once_t::lock is used to synchronize tasks (taken by one
task and given by the other) so it should not be a mutex.
Convert to semaphore, and rename to indicate its purpose.
A problem if the Python interpreter used for idf.py (or set via PYTHON
variable) didn't match
"/usr/bin/env python" (or the associated executable for .py files, on
Windows).
Closes https://github.com/espressif/esp-idf/issues/3160
Possibly also fix for https://github.com/espressif/esp-idf/issues/2936
Adds build system test to catch any future direct execution of Python in
the standard build process.
