Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | Add "#ifdef CONFIG_ARCH_IRQPRIO" where needed | Anton Babushkin | 2014-08-14 | 4 | -0/+8 |
| | |||||
* | STM32: Handle setting of USART CR1_M when 8 bits of data plus parity | Gregory Nutt | 2014-07-23 | 1 | -19/+27 |
| | |||||
* | stm32_i2c: handle more unexpected ISR conditions | Andrew Tridgell | 2014-07-12 | 1 | -1/+11 |
| | | | | | | | | this clears the interrupt for more unexpected ISR conditions that otherwise lead to an ISR storm. When that happens no user threads are able to run as the I2C ISR is continually raised Thanks to Lorenz for help in tracking this down | ||||
* | stm32_i2c: check for msgc underflow in I2C ISR | Andrew Tridgell | 2014-07-12 | 1 | -17/+24 |
| | | | | | | a spurious I2C interrupt could cause priv->msgc to underflow, leading to memory being overwritten. This seems to have been the cause of at least some of lockups we have seen | ||||
* | stm32_i2c: ensure we don't change priv-> after semaphore is released | Andrew Tridgell | 2014-07-12 | 1 | -1/+2 |
| | | | | | | | this prevents a race condition where a subsequent I2C transfer could be lost Many thanks to hxxnrx@gmail.com for reporting this bug | ||||
* | stm32_i2c: fixed timeout handling | Andrew Tridgell | 2014-07-12 | 1 | -2/+2 |
| | | | | thanks to hxxnrx@gmail.com for noticing this | ||||
* | Revert "Revert "stm32_i2c: ensure that unexpected ISR can't overwrite stale ↵ | Lorenz Meier | 2014-07-12 | 1 | -1/+10 |
| | | | | | | data"" This reverts commit 7a7a07ed0ec492b3ab9f87467189886a66b2fc1a. | ||||
* | Fix for sed "permission denied" error in Windows build | Kynos | 2014-05-24 | 1 | -1/+2 |
| | | | | | Cygwin sed in-place edit (option -i) broken, will mess up permissions. Changed to work without in-place edit. | ||||
* | mtd driver: add missing SPI header | Lorenz Meier | 2014-05-20 | 1 | -1/+1 |
| | |||||
* | stm32: add sw workaround for broken stm32 hw RTS implementation | Kevin Hester | 2014-05-20 | 3 | -7/+134 |
| | | | | | | | | | | | | | | adds support for manually asserting nRTS when the OS buffer is nearly full and deasserting it when it has room again. This makes the RTS pin actually work in a useful way (the stm32 hw implementation is broken - it asserts nRTS as soon as more than one character is received). We assert nRTS from inside of up_recvchars. For deassertion we needed a new callback from the general serial device layer when characters have been dequed. onrxdeque was added to uart_ops_s for this purpose and implemented (currently) only on the stm32. | ||||
* | Disable -Wshadow warnings for system headers - alternatively we could use ↵ | Lorenz Meier | 2014-05-16 | 4 | -0/+12 |
| | | | | #pragma GCC system_header in those | ||||
* | include/cxx: Fix some bad idempotence defintions in header files | Lorenz Meier | 2014-05-16 | 1 | -4/+4 |
| | |||||
* | Revert "stm32_i2c: ensure that unexpected ISR can't overwrite stale data" | Lorenz Meier | 2014-05-14 | 1 | -10/+1 |
| | | | | | | This reverts commit 099b2017ccfa0fc7696f168db2a0c2e13fe5edbc. This is only a temporary measure to gain the time to apply a more complete fix without side effects. | ||||
* | Merge pull request #8 from sarthakkaingade/master | Lorenz Meier | 2014-05-07 | 1 | -1/+1 |
|\ | | | | | Corrected pinmaps for stm32f40xxx processor | ||||
| * | Merge https://github.com/PX4/NuttX | Sarthak Kaingade | 2014-02-24 | 8 | -95/+650 |
| |\ | |||||
| * | | Corrected pinmaps for stm32f40xxx processor | Sarthak Kaingade | 2013-10-28 | 1 | -1/+1 |
| | | | |||||
* | | | Declared log2l() in ARM math.h | Pavel Kirienko | 2014-05-07 | 1 | -0/+1 |
| | | | |||||
* | | | Fixed include guard in cstdio. Broken include guard was preventing cstdio to ↵ | Pavel Kirienko | 2014-05-05 | 1 | -3/+3 |
| | | | | | | | | | | | | compile if cctype was included first (and vice versa); as a result, the code that needed both cctype and cstdio was failing to compile. | ||||
* | | | stm32_i2c: ensure that unexpected ISR can't overwrite stale data | Andrew Tridgell | 2014-04-28 | 1 | -1/+10 |
| | | | | | | | | | | | | | | | | | | | | | | | | if we get an ISR leftover from a previous read transfer it could use stale values for priv->dcnt and priv->ptr to overwrite stack that is no longer in scope. This resets dcnt and ptr on each new stm32_i2c_process() call to ensure that any unexpected data bytes are discarded | ||||
* | | | FAT: move cluster expansion checks to start of IO loops | Andrew Tridgell | 2014-03-04 | 1 | -54/+50 |
| |/ |/| | | | | | | | | | | | this fixes a filesystem corruption bug at cluster boundaries. Thanks to tkloss for reporting the issue. Fixes issue #12 | ||||
* | | stm32_serial: fixed spelling of FLOWCONTROL config option in stm32 serial code | Andrew Tridgell | 2014-02-10 | 1 | -20/+20 |
| | | | | | | | | this fixes flow control on pixhawk | ||||
* | | typo, wide branch should be narrow | none | 2014-01-25 | 1 | -1/+1 |
| | | |||||
* | | Assembler changes with the gcc-47 distribution from ARM mean that we need to ↵ | none | 2014-01-22 | 1 | -16/+16 |
| | | | | | | | | be explicit about branch sizes; one or more of the wide branch opcodes results in bad table branching. | ||||
* | | Merge pull request #9 from PX4/STM32F1_singlewire_uart | Lorenz Meier | 2014-01-21 | 1 | -0/+13 |
|\ \ | | | | | | | STM32F10xxx single wire usart support | ||||
| * | | small whitespace correction | Thomas Gubler | 2014-01-02 | 1 | -1/+2 |
| | | | |||||
| * | | STM32F10xxx single wire usart support | Thomas Gubler | 2014-01-02 | 1 | -2/+14 |
| | | | |||||
* | | | More compile fixes for RAMTRON device | Lorenz Meier | 2014-01-16 | 1 | -0/+1 |
| | | | |||||
* | | | Ramtron compile fix | Lorenz Meier | 2014-01-16 | 1 | -1/+1 |
| | | | |||||
* | | | Range checking on clock argument for RAMTRON set speed IOCTL | Lorenz Meier | 2014-01-16 | 1 | -4/+11 |
| | | | |||||
* | | | mtd: added MTDIOC_SETSPEED ioctl | Andrew Tridgell | 2014-01-16 | 2 | -20/+23 |
| | | | | | | | | | | | | allows setting of SPI bus speed | ||||
* | | | mtd: added timeouts to at24xx eeprom driver | Andrew Tridgell | 2014-01-16 | 1 | -2/+13 |
| | | | |||||
* | | | ramtron: the driver has been tested now | Andrew Tridgell | 2014-01-16 | 1 | -1/+1 |
| | | | |||||
* | | | ramtron: give errors for IO operations immediately | Andrew Tridgell | 2014-01-16 | 1 | -24/+27 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | waiting for write completion and checking the status register on read will improve the chance that we catch IO errors immediately rather than on the next IO. This improves the chance that callers will be able to take reasonable action on IO errors. We have one example board that does give intermittent IO errors, and this change helps for that board | ||||
* | | | Abort after N retries, propagate error throughout driver layers | Lorenz Meier | 2014-01-15 | 1 | -11/+37 |
| | | | |||||
* | | | FAT: fixed start cluster of .. directory entry | Andrew Tridgell | 2014-01-14 | 1 | -1/+5 |
| | | | | | | | | | | | | this should be 0 on FAT32 as well (tested on Windows8 and Linux) | ||||
* | | | FAT: cope with files of size equal to multiple of cluster size | Andrew Tridgell | 2014-01-14 | 1 | -2/+2 |
|/ / | | | | | | | when the file size is 4096 we should use 1 cluster, not two | ||||
* | | Merge branch 'master' of github.com:PX4/NuttX | Lorenz Meier | 2013-12-17 | 1 | -1/+1 |
|\ \ | |||||
| * | | Minor adjustment to make CLANG assembler happy | Lorenz Meier | 2013-12-10 | 1 | -1/+1 |
| | | | |||||
* | | | Fixed location of DMA console reinitialization | Lorenz Meier | 2013-12-17 | 1 | -3/+2 |
|/ / | |||||
* / | Fixes for sim for CLANG | Lorenz Meier | 2013-12-09 | 2 | -6/+494 |
|/ | |||||
* | strncpy bug fixed: fill tail of dest with zeroes if src size < n | Anton Babushkin | 2013-10-23 | 1 | -0/+1 |
| | |||||
* | sdio: fixed SDIO_DMAPREFLIGHT() macro to return fault code | Andrew Tridgell | 2013-10-10 | 1 | -3/+2 |
| | | | | the logical && was swallowing the -EFAULT return | ||||
* | Fix a typo that disconnected the DMA range validation checker from the ↵ | px4dev | 2013-09-29 | 1 | -1/+1 |
| | | | | | | actual DMA path. Now SDIO DMA ranges are properly checked, and filesystem operations are correctly bounced when they cannot be DMA'ed. | ||||
* | Also consider negative values to be valid, and while we are at it add ↵ | Lorenz Meier | 2013-09-22 | 1 | -2/+9 |
| | | | | support for the full valid C99 number symbol range plus the binary numbers NuttX supports | ||||
* | Also consider negative values to be valid | Lorenz Meier | 2013-09-22 | 1 | -2/+5 |
| | |||||
* | Cover a range of sscanf() corner cases, including %n not being the last ↵ | Lorenz Meier | 2013-09-22 | 1 | -17/+82 |
| | | | | format specifier and trailing characters after a successful number pattern. Behaves mostly now as the POSIX variant, with all tested non-error cases being exactly the same. In particular for %n the number of returned characters now matches correctly in presence of trailing whitespace. | ||||
* | Merge branch 'master' of github.com:PX4/NuttX | Lorenz Meier | 2013-09-21 | 10 | -100/+524 |
|\ | |||||
| * | Missed de-optioning the DMA preflight check. | px4dev | 2013-09-11 | 1 | -2/+0 |
| | | |||||
| * | Add an explicit preflight call for SDIO DMA operations, since we can't error ↵ | px4dev | 2013-09-08 | 3 | -29/+138 |
| | | | | | | | | out of the DMA setup (the card is already committed at that point). | ||||
| * | Fix DMA channel definitions for UART7/8 | px4dev | 2013-09-01 | 1 | -4/+4 |
| | |