| 00:06.42 | ManoftheSea | mnemoc: thanks. Who are you? I'll look for him, next midnight-thirty. |
| 02:12.08 | ManoftheSea | Hmm... So, to my eyes, sw_ahci_platform is entirely redundant code. |
| 02:12.24 | ManoftheSea | It's a direct copy of ahci_platform, with function names change. |
| 08:03.16 | *** join/#arm-netbook gordan (~gordan@78-86-195-86.zone2.bethere.co.uk) |
| 09:38.13 | *** join/#arm-netbook gordan (~gordan@178.102.245.219) |
| 11:26.00 | mnemoc | ManoftheSea: me? just a c programmer living in spain wanting to gain some new skills |
| 12:01.00 | *** join/#arm-netbook gordan (~gordan@178.102.245.219) |
| 12:01.34 | ManoftheSea | cool. |
| 12:03.37 | ManoftheSea | are you familiar with volatile and bitbanging? |
| 12:03.56 | ManoftheSea | there's a section of code that seems unnecessarily complex. |
| 12:04.19 | ManoftheSea | in drivers/ata/sw_ahci_platform.c, the function sw_ahci_phy_init |
| 12:04.32 | ManoftheSea | basically, it declares a variable tmp |
| 12:04.57 | ManoftheSea | it reads from I/O mapped memory, which is declared volatile |
| 12:05.21 | ManoftheSea | Then it |= and &= ~(x) |
| 12:06.09 | ManoftheSea | well, sometimes it's doing "tmp |= 0x3<<6 |
| 12:06.38 | ManoftheSea | and other times it turns off bits, only to turn them right back on again in the next line. |
| 12:07.12 | ManoftheSea | if you assign a volatile to a var on the stack, you're not writing all those ops to the i/o memory, right? |
| 12:07.28 | ManoftheSea | not until you write back to the volatile int * |
| 12:08.28 | ManoftheSea | second question... would a for loop that does nothing get optimized out? Or is kernel code not optimized like that? |
| 14:05.46 | *** join/#arm-netbook gordan (~gordan@178.102.245.219) |
| 14:23.52 | mnemoc | ManoftheSea: do-nothing code is eliminated unless you -O0. you can use objdump to see the final assembly and confirm |
| 14:25.20 | mnemoc | ManoftheSea: I'm still waiting for a sun4i tablet which left china on dec 12 to be able to really test the effects of using gcc 4.6 and polishing the code for mainlining |
| 14:26.30 | mnemoc | ManoftheSea: without hardware it's kind of pointless to start the cleanup... we don't even know if the code in the repo works at all |
| 14:35.28 | *** join/#arm-netbook gordan_ (~gordan@178.102.245.219) |
| 18:40.25 | *** join/#arm-netbook gordan (~gordan@78-86-195-86.zone2.bethere.co.uk) |
| 23:55.49 | ManoftheSea | mnemoc: well sure... but I'm trying to learn the kernel so as to be more useful later. |