On Sat, Feb 20, 2016 at 03:20:15PM +0100, Gerd Hoffmann wrote:
Use case: cf cards behind sata-ide bridge, which might not support the default transfer mode.
Based on a patch by Werner Zeh <werner.zeh@siemens.com>, with some minor tweaks applied.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com> --- src/hw/ahci.c | 58 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ src/hw/ata.h | 5 +++++ 2 files changed, 63 insertions(+)
diff --git a/src/hw/ahci.c b/src/hw/ahci.c index 9310850..ea862f2 100644 --- a/src/hw/ahci.c +++ b/src/hw/ahci.c @@ -515,6 +515,64 @@ static int ahci_port_setup(struct ahci_port_s *port) , ata_extract_version(buffer) , (u32)adjsize, adjprefix); port->prio = bootprio_find_ata_device(ctrl->pci_tmp, pnr, 0); + + s8 multi_dma = -1; + s8 pio_mode = -1; + s8 udma_mode = -1; + // If bit 2 in word 53 is set, udma information is valid in word 88. + if (buffer[53] & 0x04) { + udma_mode = 6; + while ((udma_mode >= 0) && + !((buffer[88] & 0x7f) & ( 1 << udma_mode ))) { + udma_mode--; + } + } + // If bit 1 in word 53 is set, multiword-dma and advanced pio modes + // are available in words 63 and 64. + if (buffer[53] & 0x02) { + pio_mode = 4; + multi_dma = 3; + while ((multi_dma >= 0) && + !((buffer[63] & 0x7) & ( 1 << multi_dma ))) { + multi_dma--; + }
This indentation should be fixed. Otherwise, looks good to me. -Kevin