--- Log opened Sat May 21 00:29:43 2011
00:29 -!- Keripo [~Keripo@c-76-28-198-27.hsd1.wa.comcast.net] has joined #freemyipod
01:02 -!- k1ddyt3ch [~techno@host86-147-48-126.range86-147.btcentralplus.com] has left #freemyipod
01:06 -!- kleemajo [~kleemajo@host176-169.resnet.ubc.ca] has quit [Remote host closed the connection]
02:21 -!- [7] [~TheSeven@rockbox/developer/TheSeven] has quit [Disconnected by services]
02:21 -!- TheSeven [~TheSeven@rockbox/developer/TheSeven] has joined #freemyipod
04:19 -!- posixninja [~textual@h177.56.23.98.dynamic.ip.windstream.net] has quit [Quit: Computer has gone to sleep.]
05:00 -!- clustur [~logger@c-98-249-104-118.hsd1.tn.comcast.net] has quit [Remote host closed the connection]
05:00 -!- clustur [~logger@c-98-249-104-118.hsd1.tn.comcast.net] has joined #freemyipod
05:48 -!- S_a_i_n_t [~st.lasciv@124-197-3-117.callplus.net.nz] has quit [Ping timeout: 276 seconds]
05:52 -!- [Saint] [~st.lasciv@124-197-3-117.callplus.net.nz] has joined #freemyipod
07:15 -!- n1s [~quassel@rockbox/developer/n1s] has joined #freemyipod
08:19 -!- n1s [~quassel@rockbox/developer/n1s] has quit [Remote host closed the connection]
10:01 -!- hiddeboomsma [~hiddeboom@lkt.hostnet.nl] has left #freemyipod
11:00 -!- clustur [~logger@c-98-249-104-118.hsd1.tn.comcast.net] has quit [Remote host closed the connection]
11:00 -!- clustur [~logger@c-98-249-104-118.hsd1.tn.comcast.net] has joined #freemyipod
11:36 -!- Keripo [~Keripo@c-76-28-198-27.hsd1.wa.comcast.net] has quit [Quit: Leaving.]
11:39 -!- benedikt93 [~benedikt9@unaffiliated/benedikt93] has joined #freemyipod
13:14 -!- liar [~liar@clnet-p09-185.ikbnet.co.at] has joined #freemyipod
16:02 < TheSeven> user890104: just read your beeper source code
16:02 < TheSeven> copying the song to the stack seems to be a bad idea to me
16:06 < user890104> what's your suggestion?
16:07 < user890104> i don't really have much expreience in embedded software development, i would do it that way if it runs on a computer
16:16 -!- user890104 [Venci@venci-notebook-lan.ipv6.6bez10.info] has quit []
17:00 -!- clustur [~logger@c-98-249-104-118.hsd1.tn.comcast.net] has quit [Remote host closed the connection]
17:00 -!- clustur [~logger@c-98-249-104-118.hsd1.tn.comcast.net] has joined #freemyipod
17:32 -!- [Saint] [~st.lasciv@124-197-3-117.callplus.net.nz] has quit [Disconnected by services]
17:32 -!- S_a_i_n_t [~st.lasciv@124-197-3-117.callplus.net.nz] has joined #freemyipod
17:32 -!- benedikt93_ [~benedikt9@p5B0C47BB.dip.t-dialin.net] has joined #freemyipod
17:32 -!- benedikt93 [~benedikt9@unaffiliated/benedikt93] has quit [Disconnected by services]
17:33 -!- benedikt93_ is now known as benedikt93
17:33 -!- benedikt93 [~benedikt9@p5B0C47BB.dip.t-dialin.net] has quit [Changing host]
17:33 -!- benedikt93 [~benedikt9@unaffiliated/benedikt93] has joined #freemyipod
17:36 -!- Keripo [~Keripo@c-76-28-198-27.hsd1.wa.comcast.net] has joined #freemyipod
17:40 -!- Keripo [~Keripo@c-76-28-198-27.hsd1.wa.comcast.net] has quit [Ping timeout: 246 seconds]
18:10 -!- posixninja [~textual@h177.56.23.98.dynamic.ip.windstream.net] has joined #freemyipod
19:19 -!- user890104 [~Venci@212.233.188.54] has joined #freemyipod
19:23 < yuriks> benedikt93: is the nano3 sdram working already? (I'm not the same one who asked the other day, but also interested)
19:24 < benedikt93> yuriks, user890104 
19:24 < benedikt93> ...
19:25 < benedikt93> damn, I didn't yet want to hit enter
19:25 < yuriks> yeah?
19:25 < yuriks> oh, xD
19:26 < benedikt93> fact is, that the i2c part of it never worked because I somehow failed at imitating apple's code, but I might go through this again (I never actually needed sdram, so I didn't reallly bother)
19:27 < benedikt93> TheSeven, or did you fix it in the meantime?
19:27 < benedikt93> I might also just copy&paste what TheSeven wrote for emcore on the classics
19:28 < yuriks> he said he had no idea if it worked (asked me to ask you)
19:29  * benedikt93 thought for some time that his sdram init *did* work, as it apparently is enough to setup the memory controller when the ipod only was in standby before
19:29 < benedikt93> I might just fix the i2c part first
19:30 < yuriks> ah, so it isn't a complete setup, just for wake up?
19:31 < benedikt93> it is complete, but when booting from standby, it doesn't matter when the code setting up the pmu fails, as the sdram seems to still be powered at that time
19:31 < benedikt93> (the pmu is connected by i2c)
19:32 < yuriks> ah, I see
19:33 < benedikt93> you own a Nano 3G, too?
19:33 < yuriks> yeah
19:33 < yuriks> I'm free for any testing
19:33 < yuriks> (that doesn't involve bricking or wiping it out too routinely :)
19:33 < user890104> i have a nano3g, too
19:33 < user890104> that's why i asked
19:35 < yuriks> is the official ARM ABI the one used for most things?
19:35 < yuriks> (in the ARM world, no ipod specifically)
19:35 < yuriks> not*
19:44 < TheSeven> nowadays probably the EABI
19:46 < yuriks> I'm doing some bare arm coding right now and am mostly just making up my own stuff based from my experience with x86 :P, specially with regards to the stack
19:46 < yuriks> results are probably being horrible
20:16 < TheSeven> yuriks: can you show me some example?
20:16 < TheSeven> if there are some misunderstandings about the platform, i could probably point you to those if I see your code
20:22 < user890104> TheSeven: why have you used memalign instead of malloc when allocating memory to fit a file from libboot
20:22 < user890104> and do i need to use it for allocating my arrays of integers?
20:23 < TheSeven> you should use memalign for all buffers that do any kind of I/O
20:24 < TheSeven> otherwise the I/O apis may either fail to work properly (intermittent failures, really nasty), reject the request, panic, or do copying to an aligned buffer internally
20:25 < TheSeven> this does not apply to all I/O, but to everything that uses DMA in one or another way
20:25 < TheSeven> that's USB, SPI, NAND, ATA, MMC, I2S and probably some more things
20:25 < TheSeven> I2C is the only thing that comes to my mind that doesn't do DMA
20:27 -!- user890104 [~Venci@212.233.188.54] has quit [Remote host closed the connection]
20:27 < yuriks> TheSeven: http://pastebin.com/eqRuSpMW http://pastebin.com/b4M5yftw http://pastebin.com/t8C0jarD
20:27 < yuriks> that's some of the code I've wrote
20:28 < yuriks> I'm mainly not sure how to do stack variables
20:29 < TheSeven> heh
20:29  * TheSeven got used to -fomit-frame-pointer
20:29 < TheSeven> i can't even remember how the frame pointer is usually used on ARM
20:29 -!- liar [~liar@clnet-p09-185.ikbnet.co.at] has quit [Ping timeout: 258 seconds]
20:29 -!- user890104 [~Venci@6bez10.info] has joined #freemyipod
20:29 < TheSeven> in fact i've never seen it being used in any code that i've disassembled
20:30 < TheSeven> and substracting #11 from the stack pointer doesn't seem like a good idea
20:30 < TheSeven> you should align that to the next 4-byte boundary
20:30 < yuriks> oh, you're right
20:31 < yuriks> stupid emulators, not crashing on me :P
20:31 -!- liar [~liar@clnet-p09-185.ikbnet.co.at] has joined #freemyipod
20:32 < TheSeven> that won't neccessarily crash, but i might do very weird things(tm)
20:32 < TheSeven> things will start wrapping around
20:32 < yuriks> yeah
20:32 < yuriks> it's probably fine here cause I'm not using the stack for anything else
20:32 < yuriks> (I'm saving register on it, but I'm not using those regs later anyway)
20:34 < yuriks> (fine? I meant 'working' :P)
20:34 < TheSeven> anyway, the code seems to be ABI compatible
20:35 < TheSeven> however you're doing some things in a weird way
20:35 < yuriks> like what?
20:35 < TheSeven> usually LR is saved on entry, used as a scratch register in the function (as function calls will corrupt it anyway) and then restored into PC on exit
20:36 < TheSeven> you're leaving LR alone, not calling any subfunctions, and allocating some other scratch register instead
20:36 < TheSeven> this is of course compatible, but not the way things are usually done
20:36 < yuriks> oh, that's a nice trick
20:36 < TheSeven> beware that this trick breaks thumb compatibility on older ARM cores
20:37 < yuriks> this is gonna run on a custom ARM-compatible cpu without thumb support anyway :P
20:37 < TheSeven> (you need a BX instruction to properly return to/from thumb mode on them, and an LDMxy won't work)
20:37 < TheSeven> ok, no need to care about that then
20:38 < TheSeven> another thing that's just a convention and doesn't even affect the output binary is the LDM/STM notation you're using
20:39 < TheSeven> DA/DB/IA/IB are usually used for general purpose read/write operations, and the equivalent EA/ED/FA/FD mnemonics are used for stack operations
20:39 < TheSeven> that's just to make the code more readable
20:39 < TheSeven> STMFD == STMDB, LDMFD == LDMIA
20:40 < yuriks> hmm
20:40 < TheSeven> so you can use the same suffix everywhere you're accessing the stack, depending only on the stack type
20:40 < TheSeven> and you'll usually use an FD stack (full descending)
20:41 < yuriks> apparently there are push/pop mnemonics too (for stmfd/ldmfd)
20:41 < TheSeven> yeah, those are actually thumb mnemonics, which are usually translated by the assembler when writing ARM mode code
20:43 < yuriks> thanks for the pointers :)
20:44 < TheSeven> but yeah, you're doing really good so far :)
20:45 < TheSeven> it seems to me like you have never written ARM code before, but know quite a lot of ARM theory (have read their ARM/TRMs) and have assembly programming background from some other architecture
20:45 < yuriks> and that would be correct
20:46 < yuriks> I know x86, a tiny bit of 68k. And have read the architecture spec (and this one http://nocash.emubase.de/gbatek.htm#cpuoverview) a bunch of times
20:56 < fmibot_> New commit by user890104 (r715): Beeper: optimize memory allocation a bit
20:57 < fmibot_> r715 build result: emcore: All green!
20:58 < TheSeven> user890104: why are you copying this stuff around at all?
20:59 < TheSeven> i can't see any reason why you couldn't play directly from the input buffer
21:09 < Farthen> TheSeven: does ++i in a for loop have any impact? shouldn't it be i++?
21:09 < yuriks> it increments i
21:09 < Farthen> sure but does it have any impact if you prefix ++ in a for loop?
21:09 < TheSeven> ++i == pre-increment, i++ == post-increment
21:09 < Farthen> yes, sure
21:09 < yuriks> outside of a more complex expression ++i and i++, except for i++ being slower if the compiler is dumb
21:09 < Farthen> again, does that have any impact if used in a for loop?
21:09 < yuriks> are the same*
21:10 < Farthen> ok
21:10 < TheSeven> if you mean the third argument of a for statement, it doesn't matter (personal preference)
21:10 < TheSeven> in some other places it does
21:10 < Farthen> ok
21:10 < Farthen> i was referring to user890104's code and i have never seen ++i in a for loop yet, that's why i was asking
21:11 < yuriks> it's pretty common in C++ code
21:12 < yuriks> (where i may be an iterator instead of a simple int, and thus i++ may have different semantics (if the library writer hates you) than ++i
21:15 < user890104> TheSeven: i don't know how to use the data in the form i loaded it so i split it to two arrays
21:16 < benedikt93> you  might access it by some simple struct
21:17 < TheSeven> either that, or just by the pointers that you used to copy it into the array
21:17 < benedikt93> struct tone { uint8_t cycles, uint8_t length }
21:18 < benedikt93> and then cast the file pointer to a pointer to such struct
21:18 < Farthen> IMHO that would be the most clean solution to that
21:18 < Farthen> because it makes the following code easy to understand
21:19 < user890104> yeah, i'll try it then
21:19 < yuriks> is the status page on the wiki up-to-date re. the 3G?
21:19 < Farthen> or rather a pointer to an array of such structs, wouldn't it?
21:20 < benedikt93> wouldn't that basically be the same?
21:20 < yuriks> one extra level of indirection
21:20 < yuriks> that's not really needed...
21:21 < Farthen> i don't know but you may be right, yes
21:21 < Farthen> difference would be that you would need to increment the pointer by sizeof(struct) every time
21:22 < Farthen> which my be slower than using an array
21:22 < Farthen> (because it has to do sizeof every time)
21:22 < TheSeven> eh?
21:22 < yuriks> but you need the sizeof for the array too...
21:22 < Farthen> whops
21:22 < benedikt93> I thought it the other way around, beacause it would multiply always when using an array
21:22 < TheSeven> the difference would be an extra pointer on the stack and an extra dereference
21:23 < yuriks> benedikt93: that's free also
21:23 < yuriks> since the multiply is just a shift
21:23 < TheSeven> Farthen: you can just pass arrays as arguments in C
21:23 < benedikt93> <yuriks> is the status page on the wiki up-to-date re. the 3G? <-- yep
21:23 < TheSeven> yuriks: nah, if it's clever, the compiler can completely eliminate it
21:24 < Farthen> TheSeven: yes? what was the point of you telling me that?
21:24 < TheSeven> Farthen: and sizeof is not a function, but a compile-time evaluated constant
21:24 < yuriks> TheSeven: oh, yeah, but even if it doesn't, that kind of base + index * power_of_2 sort of adressing is free on ARM
21:25 < TheSeven> assuming r0 = data pointer, r1 = size:
21:25 < TheSeven> loop:
21:25 < yuriks> though I guess you use an extra register
21:26 < TheSeven> ldmia r0!, {r2, r3}
21:26 < TheSeven> <do something>
21:26 < TheSeven> subs r1, r1, #8
21:26 < TheSeven> bne loop
21:26 < yuriks> oh, multiple load
21:26 < TheSeven> well, or two ldr rX, [r0], #4 instructions
21:27 < yuriks> yeah
21:55 -!- benedikt93 [~benedikt9@unaffiliated/benedikt93] has quit [Quit: "Facts do not cease to exist because they are ignored." - Aldous Huxley]
22:02 -!- user890104 [~Venci@6bez10.info] has quit [Ping timeout: 252 seconds]
22:34 -!- liar [~liar@clnet-p09-185.ikbnet.co.at] has quit [Ping timeout: 258 seconds]
22:36 -!- user890104 [Venci@venci-notebook-lan.ipv6.6bez10.info] has joined #freemyipod
22:51 < fmibot_> New commit by user890104 (r716): Beeper: use less memory
22:51 < fmibot_> r716 build result: emcore: All green!
22:52  * user890104 thinks that his app is pretty much complete for now
22:53 < user890104> i'm thinking about writing a file manager that runs on the device itself
22:57 -!- liar [~liar@clnet-p09-185.ikbnet.co.at] has joined #freemyipod
23:00 -!- clustur [~logger@c-98-249-104-118.hsd1.tn.comcast.net] has quit [Remote host closed the connection]
23:00 < yuriks> heh, I still remember how to run code on my 3g :P
23:00 -!- clustur [~logger@c-98-249-104-118.hsd1.tn.comcast.net] has joined #freemyipod
23:28 -!- bairchan [~boromir@pool-72-70-187-188.hrbgpa.fios.verizon.net] has quit [Ping timeout: 240 seconds]
23:29 -!- bairchan [~boromir@pool-72-70-187-188.hrbgpa.fios.verizon.net] has joined #freemyipod
23:47 < yuriks> is there a way to tell IDA to treat unknown data as dwords instead of bytes? (Or to mass tag some memory as being dwords)
--- Log closed Sun May 22 00:08:07 2011