On 3/25/10 5:18 PM, Myles Watson wrote:
On Thu, Mar 25, 2010 at 7:56 AM, Myles Watson mylesgw@gmail.com wrote:
It seems like others must have this problem of needing to force a binary blob into a struct.
Hi, I've only been following this at a distance. I assume this all happens after memory is turned on. Why not just have a function to fill the struct given the binary blob? I'm starting to worry about all the tricks people want to play with gcc.
Memory is on. The blob is in flash, and the first bit of the blob is the header with the length field.
We do a memcpy from flash to the struct,
Actually this is a memcpy from memory to the cbmem area, but after it's copied it's not treated as a struct anymore.
but first we have to cast it to a header so we can read out the length to copy.
Which makes me think if there are not other ways to determine the size of an array. Maybe sizeof(AmlCode) ?
Gcc doesn't like the cast, so we just declare that we have the header stored in flash.
One way to solve it would be to copy the header first, then read the length and copy the rest...
That's more copying than my variant with the extra pointer. Since we copied (or uncompressed) the dsdt from flash to ram already before, and are going to make another copy, I suggest rather keeping an extra 4 bytes, than copying the complete header for no apparent reason...
Stefan