Skip to contentSkip to conversation

New kiobuf diffs for 2.3.99-pre9-2

PATCH New Stephen C. Tweedie 1 participant 1 message May 18, 2000 Memory management

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 New kiobuf diffs for 2.3.99-pre9-2

Hi all,

I've batched the current kiobuf code up at

ftp://ftp.uk.linux.org/pub/linux/sct/fs/raw-io/kiobuf.2.3.99.pre9-2.tar.gz

There are a couple of new things in here since the last version. In particular, it includes Kanoj Sarcar's fork fixes and Dave Miller's pipe speedup code, as well as a few minor bug fixes and fixes to error return codes.

I'll wait for feedback and then file those bits which are clearly bugfixes on to Linus. The more significant changes will probably be post-2.4 items.

Quoted 1 line

From the README:

This tarball contains the following patches, to be applied in order:

01-mapfix.diff

map_user_kiobuf() retries failed maps to cover a race in which the swapper steals a page before the kiobuf has grabbed and locked it.

02-iocount.diff

Kanoj Sarcar's fixes to allow kiobufs to work properly over fork(), even on threaded applications.

03-davem-pipe.diff

Dave Miller's rocking pipe code using kiobufs for a 2* throughput improvement on simple streaming pipe I/O.

04-eiofix.diff

Fix to return -EIO instead of 0 if a raw I/O read or write encounters an error in the first block.

05-kvmap.diff

New code to allow:

1) map_kernel_kiobuf: the analogue of map_user_kiobuf,
except that it works on kernel virtual addresses instead. Even vmalloc()ed regions work.

2) Add a "flags" argument to map_*_kiobuf. The only flag honoured is

MAP_PRIVATE: any mappings of the kiobuf will be kept
process-local over forks. Without this, the pages will remain shared over fork (which will cause real problems if you map the pages into a MAP_PRIVATE vma in user space).

MAP_PRIVATE is used by the raw character device.

3) Add kvmap infrastructure to allow mmap() of any kiobuf. Includes a sample driver in Documentation/kiobuf.sample.c to show how it can work.

06-enxio.diff

Return ENXIO on read/write at or beyond the end of the device for raw I/O

--Stephen Tweedie <[email protected]>

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/