[PATCH] jffs2 summary allocation

[David Brownell]

On Friday 04 April 2008, Andrew Morton wrote:
> This problem comes up pretty often. 

Which problem -- kmalloc(BIG)?  Or dma(dma-unsafe-mem)?
Or something else?

The specific issue here might best be described as JFFS2
making a bad assumption:  that MTD drivers never use DMA.
The I2C stack does that in some cases (for example, I/O
buffers in i2c_smbus_xfer_emulated are on-stack), but I'd
not call that an especially common assumption.


> Rather than open-coding it yet again 
> it'd be nice to have a little bit of library code which manages an array of
> pages and which has accessors for common operations like
> read/write-u8/u16/u32/u64, memset, memcpy, etc.

If array-of-pages is to be more widely adopted, that'd
make sense.  The MTD framework is a bit odd in that
respect ... it has block devices but doesn't use the
scatterlist primitives.

- Dave