Hi,
We have an system where we take care of buffering the file data while
reading and writing. The buffer used for the same are reserved memory which is
not visible to kernel. (Kseg1 address in MIPS processor and is reserved by
specifying an offseted address while booting). The data is filled by the
hardware in the system. Now we need to write the buffer into a file. The data
is a huge about 1MB of buffers. I dont want to split the write/read request to
block size. I also dont want it be page cached/buffer cached before going to
the IDE driver.
PS: Only few files are of this type, other files are generated using the normal
malloced buffers.
In order to achive this, i think the following changes are required.
1. Modification to support large block size.
2. Modification to eliminate page/buffer cache.
Is there any other way by which this can be achieved. Or can anyone give
me pointer for who to go about doing the above two modification.
Thanks and Regards
Santhosh
|