| 00:01.17 | RP_ | can anyone explain this: |
| 00:01.22 | RP_ | The w100 code hw_init function has some code if(isRemapped == 0) { remap[1-3]=ioremap(blah[1-3]); isRemapped=1; [a] } [b] |
| 00:01.27 | RP_ | If I put a printk at position [b] the driver behaves as normal. If I put it at position [a], it sulks and the chip disappears. Any ideas why? |
| 00:02.53 | CosmicPenguin | chip "disappears"? |
| 00:03.31 | RP_ | It doesn't appear to be at the io addresses it should be at |
| 00:03.48 | RP_ | and the screen does random stuff - no control over it |
| 00:04.01 | CosmicPenguin | In other words, you messed up the ioremap |
| 00:04.15 | RP_ | but I'm not changing the ioremap... |
| 00:05.06 | RP_ | I'm just changing the position of the printk to inside or outside that bracket... |
| 00:05.37 | CosmicPenguin | I told you before - its obvious that you have some tenious variables somewhere that seem to care about the size and position of your code |
| 00:05.51 | CosmicPenguin | perhaps you're doing violence to the stack, or maybe its the heap |
| 00:06.07 | CosmicPenguin | But if you're moving your code segments around, and your code is acting strange, then you have buggy code |
| 00:06.24 | RP_ | How do I find it? |
| 00:06.51 | *** join/#oe sunny (sunny@sunny.user) |
| 00:07.20 | CosmicPenguin | Well, you know that the printk seems to disturb the variables that you use for ioremap, so I would start there |
| 00:08.49 | RP_ | by looking for what? I'm really at a loss as to what I'm looking for :-/ |
| 00:09.45 | RP_ | I understand the problems of code caring about the position of things but I can't see that this does. It could well be my programming abilities just aren't up to this :-/ |
| 00:10.07 | CosmicPenguin | look at the sypmtoms, and ask yourself, how can this happen? |
| 00:10.29 | CosmicPenguin | when you try to read the ioremaped region, you get unsane values |
| 00:10.41 | CosmicPenguin | 1) you ioremaped the wrong place |
| 00:10.51 | CosmicPenguin | 2) the chip is buggy |
| 00:10.59 | CosmicPenguin | 3) you overwrote your pointer with another pointer |
| 00:11.41 | RP_ | the place I'm mapping from is hardcoded and the values the printk prints (the iomapped addresses) are always the same |
| 00:12.18 | RP_ | So that seems to rule out 1) and 3). 2) is an unknown... |
| 00:12.35 | CosmicPenguin | It doesn't look like its hard coded in your psuedo code |
| 00:13.16 | RP_ | The actual code is three of things like "remapped_base = ioremap_nocache (W100FB_PHYS_ADDRESS, REMAPPED_CFG_LEN);" |
| 00:13.36 | RP_ | where the two parameters are defines |
| 00:13.45 | emte | out of curiosity dont most system bioses have a specific memory block allocated to vidoe? |
| 00:14.24 | RP_ | emte: under i386, yes. other archs, yes/no/maybe... :) |