- 08 Dec, 2013 3 commits
-
-
Juan David González Cobas authored
-
Juan David González Cobas authored
-
Juan David González Cobas authored
We should use at least 27e6b00, to avoid the readl/writel naming conflicts addressed by that commit
-
- 22 Nov, 2013 3 commits
-
-
Tomasz Wlostowski authored
-
Tomasz Wlostowski authored
kernel: fix edge emulation timeout & check for subsequent VIC IRQs before returning from the handler. The latter keeps handling VIC irqs in a loop until none of the IRQ lines is active anymore. This prevents situations when two or more IRQs come almost simultaneously (e.g. reading sychronized sample buffers) and the one with lower priority gets starved.
-
Tomasz Wlostowski authored
-
- 18 Nov, 2013 1 commit
-
-
Tomasz Wlostowski authored
-
- 24 Oct, 2013 1 commit
-
-
Tomasz Wlostowski authored
-
- 27 Sep, 2013 10 commits
-
-
Tomasz Wlostowski authored
-
Tomasz Wlostowski authored
-
Tomasz Wlostowski authored
-
Tomasz Wlostowski authored
-
Tomasz Wlostowski authored
-
Tomasz Wlostowski authored
The VIC is the standard BE-CO-HT module for multiplexing FPGA-internal IRQ sources. Support for it in the carrier driver is provided for convenience of FMC driver developers. More info in the svec-sw manual.
-
Tomasz Wlostowski authored
-
Tomasz Wlostowski authored
-
Tomasz Wlostowski authored
-
Tomasz Wlostowski authored
-
- 28 Aug, 2013 1 commit
-
-
Juan David González Cobas authored
This way, no external dependency on /acc/... directories exists
-
- 05 Jun, 2013 5 commits
-
-
Juan David González Cobas authored
-
Juan David González Cobas authored
-
Juan David González Cobas authored
-
Juan David González Cobas authored
-
Luis Fernando Ruiz authored
-
- 27 May, 2013 1 commit
-
-
Juan David González Cobas authored
-
- 24 May, 2013 2 commits
-
-
Juan David González Cobas authored
Signed-off-by: Juan David Gonzalez Cobas <dcobas@cern.ch>
-
Juan David González Cobas authored
-
- 23 May, 2013 13 commits
-
-
Juan David González Cobas authored
-
Juan David González Cobas authored
When registering several mezzanines, the second device_register causes a reprogram that should not do anything (the bitstream should not change). The release of fmc->sdb information must happen before, so that fmc_scan_sdb_tree refills the new SDB information properly. This is a kludge over a kludge: actually, fmc_scan_sdb_tree should be called once in the carrier, so that we do not fall into this.
-
Luis Fernando Ruiz authored
-
Luis Fernando Ruiz authored
-
Luis Fernando Ruiz authored
-
Juan David González Cobas authored
-
Juan David González Cobas authored
This commit reverts 95a811a4, as the correct use of FMC_NO_MEZZANINE prevents from faking an EEPROM and registering the (physically inexistent) device
-
Juan David González Cobas authored
-
Juan David González Cobas authored
We need to track the commit this binary comes from, as a general policy
-
Juan David González Cobas authored
-
Juan David González Cobas authored
golden has the same bitbanged i2c interface as the fine delay
-
Juan David González Cobas authored
It is useless and should be no more
-
Juan David González Cobas authored
This kludge is made to cheat the fmc-bus into the belief that a valid EEPROM with no FRU information was read, given that FMC_DEVICE_NO_MEZZANINE is not used in it.
-