On Monday 06 July 2009 02:26:20 Prafulla Wadaskar wrote:
From: Mike Frysinger [mailto:vapier@gentoo.org]
On Friday 03 July 2009 13:28:01 Prafulla Wadaskar wrote:
- {
.idcode0 = MXIC_ID_MT_MX2512855E,.idcode1 = MXIC_ID_MD_MX2512855E,.page_size = 256,.pages_per_sector = 16,.sectors_per_block = 16,.nr_blocks = 256,.name = "MX25L12855E",- },
};
can you compare the code size to see if combining the id's in a u16 is better than comparing multiple u8's ?
At this moment we can save 28bytes with above suggestions, but Mx25L12805D and MX25L12855E have different memory types (0x20, 0x26 resp), and they have different protection algorithm features which I was trying to expose through "protect" command interface in my next patch, then we will need this abstraction. So I wish to keep this. What do you think?
i was referring to also the .text differences, not just the .data. so with a u16, the structure would look like: {... .id = 0x1826, .name = "MX25L12855E", ...},{... .id = 0x1820, .name = "MX25L12805D", ...},
and the probe would look something like: u16 id = idcode[0] | (idcode[1] << 8); ... if (params->id == id) ...
i dont think this would prevent you from doing protection detection on the 0x18 family as you could create a new define: #define MX_PROT_FAMILY_FOO 0x18
and then in the probe code do: if (idcode[1] == MX_PROT_FAMILY_FOO)
i dont think the u8 -> u16 conversion would cause problems here would it ? -mike