On 2/27/20 5:59 PM, Douglas Gilbert wrote:
Linux has an issue with mmap-ed multiple pages issued by alloc_pages() [with order > 0]. So when mmap(2) is called if the reserve request's scatter gather (sgat) list is either: - not big enough, or - made up of elements of order > 0 (i.e. size > PAGE_SIZE) then throw away reserve requests sgat list and rebuild it meeting those requirements. Clean up related code and stop doing mmap+indirect_io. This new mmap implementation is marginally more flexible (but still compatible with) the production driver. Previously if the user wanted a larger mmap(2) 'length' than the reserve request's size, then they needed to use ioctl(SG_SET_RESERVED_SIZE) to set the new size first. Now mmap(2) called on a sg device node will adjust to the size given by 'length' [mmap's second parameter]. Tweak some SG_LOG() levels to control the amount of debug output. Signed-off-by: Douglas Gilbert <dgilbert@xxxxxxxxxxxx> --- drivers/scsi/sg.c | 150 +++++++++++++++++++++++++++------------------- 1 file changed, 87 insertions(+), 63 deletions(-)
Reviewed-by: Hannes Reinecke <hare@xxxxxxx> Cheers, Hannes -- Dr. Hannes Reinecke Teamlead Storage & Networking hare@xxxxxxx +49 911 74053 688 SUSE Software Solutions GmbH, Maxfeldstr. 5, 90409 Nürnberg HRB 36809 (AG Nürnberg), Geschäftsführer: Felix Imendörffer
