| 00:26:14 | *** | Saving seen data "./dancer.seen" |
| 00:49:39 | | Quit othello7 (Ping timeout: 252 seconds) |
| 02:00 |
| 02:25:27 | | Join braewoods_ [0] (~braewoods@user/braewoods) |
| 02:26:18 | *** | Saving seen data "./dancer.seen" |
| 02:27:49 | | Quit braewoods (Ping timeout: 248 seconds) |
| 02:41:05 | | Join chris_s [0] (~chris_s@2a09:bac2:2a1d:1b4b::2b8:9f) |
| 03:00 |
| 03:13:20 | | Quit chris_s (Quit: Client closed) |
| 03:17:54 | | Quit floxley (Ping timeout: 248 seconds) |
| 03:18:31 | | Join braewoods__ [0] (~braewoods@user/braewoods) |
| 03:19:09 | | Quit braewoods_ (Read error: Connection reset by peer) |
| 04:00 |
| 04:03:46 | | Quit jacobk (Ping timeout: 248 seconds) |
| 04:04:51 | | Join jacobk [0] (~quassel@47-186-105-237.dlls.tx.frontiernet.net) |
| 04:26:21 | *** | Saving seen data "./dancer.seen" |
| 04:45:19 | | Quit cstine (Quit: Ping timeout (120 seconds)) |
| 04:45:38 | | Join cstine [0] (~cstine@150.136.136.191) |
| 06:00 |
| 06:26:25 | *** | Saving seen data "./dancer.seen" |
| 06:28:30 | | Join berber_l517 [0] (~berber@2a03:4000:7:4e0::) |
| 07:00 |
| 07:04:00 | | Quit yang2 (Quit: Connection closed for inactivity) |
| 07:48:19 | rb-bluebot | Build Server message: New build round started. Revision e29ddfb6be, 345 builds, 9 clients. |
| 07:48:19 | rb-bluebot | ata: Support for ATA logical sector sizes > 512B by Solomon Peachy |
| 08:00 |
| 08:00:31 | rb-bluebot | Build Server message: Build round completed after 733 seconds. |
| 08:00:33 | rb-bluebot | Build Server message: Revision e29ddfb6be result: All green |
| 08:00:41 | rb-bluebot | Build Server message: New build round started. Revision f58fad943e, 345 builds, 9 clients. |
| 08:00:41 | rb-bluebot | diskcache: Size the buffers for MAX_LOG_SECTOR_SIZE by Solomon Peachy |
| 08:13:11 | rb-bluebot | Build Server message: Build round completed after 751 seconds. |
| 08:13:13 | rb-bluebot | Build Server message: Revision f58fad943e result: All green |
| 08:13:18 | rb-bluebot | Build Server message: New build round started. Revision 9e70991174, 345 builds, 7 clients. |
| 08:13:19 | rb-bluebot | disk: Simplify LOG_SECTOR_SIZE() definition by Solomon Peachy |
| 08:26:29 | *** | Saving seen data "./dancer.seen" |
| 08:30:32 | rb-bluebot | Build Server message: Build round completed after 1034 seconds. |
| 08:30:33 | rb-bluebot | Build Server message: Revision 9e70991174 result: All green |
| 08:30:40 | rb-bluebot | Build Server message: New build round started. Revision 2156d9874f, 345 builds, 9 clients. |
| 08:30:41 | rb-bluebot | metadata: mp3: Improve support for long tags by Roman Artiukhin |
| 08:44:44 | rb-bluebot | Build Server message: Build round completed after 845 seconds. |
| 08:44:46 | rb-bluebot | Build Server message: Revision 2156d9874f result: All green |
| 08:44:53 | rb-bluebot | Build Server message: New build round started. Revision da478aaa32, 345 builds, 9 clients. |
| 08:44:53 | rb-bluebot | metadata: mp3: Simplify comment tag processing by Roman Artiukhin |
| 08:52:16 | speachy | chris_s: The plot thickens with the ipod6g weridness. I know which commit is responsible, but _why_ appears to be a deeper bug. |
| 08:54:43 | speachy | (the codepath is long successfully used on the ipod5g) |
| 08:57:54 | speachy | (albeit with a logical sector size of 2K rather than 4K) |
| 08:59:01 | rb-bluebot | Build Server message: Build round completed after 848 seconds. |
| 08:59:02 | rb-bluebot | Build Server message: Revision da478aaa32 result: All green |
| 08:59:10 | rb-bluebot | Build Server message: New build round started. Revision ee365f21c4, 345 builds, 9 clients. |
| 08:59:10 | rb-bluebot | show track info: skip Comment and codec string copies by Roman Artiukhin |
| 08:59:29 | speachy | it's debatable if some of this crap is even necessary (ie supporting ATA drives that use >512B logical sector sizes) |
| 09:00 |
| 09:07:14 | speachy | I want to finish landing the "support variable sector size" patchset first, but afterwards I'll go back and (1) rename MAX_LOG_SECTOR_SIZE/MAX_PHYS_SECTOR_SIZE to accurately reflect what they actually represent, and introduce new ones covering supporting varying-at-runtime logical/physical sector sizes. |
| 09:12:31 | rb-bluebot | Build Server message: Build round completed after 802 seconds. |
| 09:12:33 | rb-bluebot | Build Server message: Revision ee365f21c4 result: All green |
| 09:47:49 | | Quit paulk (Ping timeout: 248 seconds) |
| 09:53:59 | | Join paulk [0] (~paulk@vpn-0-22.aquilenet.fr) |
| 09:53:59 | | Quit paulk (Changing host) |
| 09:53:59 | | Join paulk [0] (~paulk@about/aquilenet/user/paulk) |
| 10:00 |
| 10:26:31 | *** | Saving seen data "./dancer.seen" |
| 10:33:31 | | Quit Tonux (Ping timeout: 252 seconds) |
| 10:33:39 | | Join Tonux_ [0] (~Tonux@193.32.127.240) |
| 10:35:30 | | Nick Tonux_ is now known as Tonux (~Tonux@193.32.127.240) |
| 10:44:16 | | Quit paulk (Quit: WeeChat 3.0) |
| 10:47:21 | | Join paulk [0] (~paulk@about/aquilenet/user/paulk) |
| 10:58:38 | | Join davisr [0] (~davisr@fsf/emeritus/davisr) |
| 11:00 |
| 11:54:48 | | Join chris_s [0] (~chris_s@2a02:26f7:ec48:4000:a112:fd8e:1145:136a) |
| 11:54:58 | rb-bluebot | Build Server message: New build round started. Revision 0386469794, 345 builds, 9 clients. |
| 11:54:58 | rb-bluebot | Fix incorrect UI font id in some plugins by Christian Soffke |
| 12:00 |
| 12:07:19 | rb-bluebot | Build Server message: Build round completed after 742 seconds. |
| 12:07:21 | rb-bluebot | Build Server message: Revision 0386469794 result: All green |
| 12:17:13 | speachy | chris_s: one unexpected thing I ran into when fixing mikmod's use of SYSFONT was that it had hardcoded assumptions about the font size. Just something to watch out for. |
| 12:19:29 | | Quit chris_s (Quit: Client closed) |
| 12:24:14 | | Join chris_s [0] (~chris_s@2a09:bac3:2918:1282::1d8:188) |
| 12:26:35 | *** | Saving seen data "./dancer.seen" |
| 12:33:10 | chris_s | thanks speachy, good point |
| 12:34:03 | chris_s | these seem to be the last vestiges of font_get being used with FONT_UI, something that was apparently deprecated.a decade ago |
| 12:34:50 | speachy | heh heh, that's fairly young by our standards |
| 12:35:07 | chris_s | :) |
| 12:40:41 | | Quit davisr (Quit: yeehaw) |
| 12:49:46 | | Join othello7 [0] (~Thunderbi@pool-100-36-176-164.washdc.fios.verizon.net) |
| 13:00 |
| 13:33:22 | | Quit chris_s (Quit: Client closed) |
| 13:41:58 | | Join JanC_ [0] (~janc@user/janc) |
| 13:41:58 | | Quit JanC (Killed (erbium.libera.chat (Nickname regained by services))) |
| 13:41:58 | | Nick JanC_ is now known as JanC (~janc@user/janc) |
| 13:42:29 | | Quit hactar|ant (Quit: and it's always been the same / it's just a complicated game) |
| 14:00 |
| 14:09:21 | | Quit othello7 (Ping timeout: 252 seconds) |
| 14:18:05 | | Quit JanC (Ping timeout: 260 seconds) |
| 14:25:32 | | Join JanC [0] (~janc@user/janc) |
| 14:26:39 | *** | Saving seen data "./dancer.seen" |
| 14:40:24 | speachy | well well well, looks like we have a smoking gun for the ipod6g weirdness. Appears to be some sort of DMA alignment issue. |
| 14:42:11 | | Join othello7 [0] (~Thunderbi@pool-100-36-176-164.washdc.fios.verizon.net) |
| 14:50:52 | | Quit othello7 (Quit: othello7) |
| 16:00 |
| 16:09:07 | | Join hactar|ant [0] (~zem@97-115-77-115.ptld.qwest.net) |
| 16:26:42 | *** | Saving seen data "./dancer.seen" |
| 16:35:00 | | Quit jacobk (Remote host closed the connection) |
| 16:44:35 | rb-bluebot | Build Server message: New build round started. Revision 9dd2781b87, 345 builds, 9 clients. |
| 16:44:36 | rb-bluebot | ipod6g: Use storage bounce buffer for ipod6g by Solomon Peachy |
| 16:45:19 | speachy | all of this crap can't be good for performance, but once it's all "correct" we can see about optimizing some of these buffers out. |
| 16:45:50 | _bilgus__ | great! hope its not actually subtle code alignment issue |
| 16:48:10 | _bilgus__ | if it is the problem will pop up at some later time in an unrelated change that moves something, we've had a few and they are usually simple if you can find it, finding it being the hard part |
| 16:49:51 | speachy | Still don't understand it, honestly. |
| 16:50:03 | _bilgus__ | like a one off error that works because it hits somebody elses padding etc |
| 16:50:20 | speachy | All I/O should be going through a 'struct sector_cache_entry' which has STORAGE_ALIGN_ATTR |
| 16:50:33 | speachy | no, I stand corrected |
| 16:50:36 | _bilgus__ | code gets aligned a bit different and now you just overwrote someones stuff |
| 16:52:00 | speachy | it only uses that on the first or last chunks of a transfer; stuff in the middle goes straight through |
| 16:52:33 | _bilgus__ | so maybe some math is still using 512? |
| 16:53:28 | speachy | oho! ipod6g doesn't define STORAGE_WANTS_ALIGN |
| 16:54:54 | | Join Everything [0] (~Everythin@178-133-1-121.mobile.vf-ua.net) |
| 16:57:26 | rb-bluebot | Build Server message: Build round completed after 771 seconds. |
| 16:57:27 | rb-bluebot | Build Server message: Revision 9dd2781b87 result: All green |
| 17:00 |
| 17:00:09 | speachy | I'll bet that's the problem. the first and last block get fixed up through the buffer but the stuff in the middle might end up unaligned. |
| 17:14:51 | _bilgus__ | I found STORAGE_ALIGN_ATTR in ata-common.c so sounds plausible to me |
| 17:16:01 | speachy | thing is ipod6g has _never_ had STORAGE_WANTS_ALIGN defined |
| 17:16:26 | _bilgus__ | there were major issues with enabling DMA from the get no? |
| 17:16:28 | speachy | so this could be a case where it accidentally worked. |
| 17:16:50 | speachy | still I'd expect a lot more to go wrong. |
| 17:16:52 | _bilgus__ | or was that one of the other ipods? |
| 17:17:01 | speachy | that was the PP-based ipods |
| 17:17:07 | speachy | DMA worked but was slower. |
| 17:17:26 | speachy | and with the iflash boards, our optimized PIO timings caused corruption. :) |
| 17:19:40 | _bilgus__ | add some buffers infront and behind fill with a sentinel and see if you can cause the issue by changing the alignment? |
| 17:24:23 | _bilgus__ | sector_cache_entry has unsigned char as first element could you put sector_t prior and get alignment? |
| 17:24:46 | speachy | I don't have one of these devices so I'm entirely dependent on 3rd parties. |
| 17:24:46 | _bilgus__ | it being an int |
| 17:24:50 | _bilgus__ | ah |
| 17:25:24 | speachy | STORAGE_ALIGN_ATTR is nil so... |
| 17:27:28 | _bilgus__ | STORAGE_WANTS_ALIGN changes that |
| 17:28:53 | speachy | yeah. I have two more test builds up, (1) is just git master, without the rest of my fat/file variable sector size stuff, and (2) swapping BOUNCE_BUFFER for WANTS_ALIGN |
| 17:36:26 | | Join Moriar [0] (~moriar@107-200-193-159.lightspeed.stlsmo.sbcglobal.net) |
| 17:42:04 | | Quit Xeha (Ping timeout: 245 seconds) |
| 17:46:23 | | Join othello7 [0] (~Thunderbi@pool-100-36-176-164.washdc.fios.verizon.net) |
| 17:58:04 | | Quit Everything (Remote host closed the connection) |
| 18:00 |
| 18:16:20 | speachy | finished my variable sector size storage rework too, properly wrapped by #ifdef(MAX_VARIABLE_LOG_SECTOR) |
| 18:17:06 | speachy | so we won't pay any significant buffer/etc size penalty. we can enable this if we ever encounter drives with >512B logical sectors. |
| 18:26:43 | *** | Saving seen data "./dancer.seen" |
| 18:52:28 | | Join _bilgus_ [0] (~bilgus@syn-162-154-213-134.res.spectrum.com) |
| 18:54:17 | | Quit _bilgus__ (Ping timeout: 255 seconds) |
| 19:00 |
| 19:43:22 | | Join massiveH [0] (~massiveH@2600:4040:a982:5400:89e:5f5f:1c66:efa7) |
| 20:00 |
| 20:26:45 | *** | Saving seen data "./dancer.seen" |
| 20:58:29 | rb-bluebot | Build Server message: New build round started. Revision 3664d382e0, 345 builds, 9 clients. |
| 20:58:30 | rb-bluebot | ata: if drive doesn't advertise a nonstandard sector size, default to 512 by Solomon Peachy |
| 21:00 |
| 21:07:27 | | Join jacobk [0] (~quassel@47-186-105-237.dlls.tx.frontiernet.net) |
| 21:11:45 | rb-bluebot | Build Server message: Build round completed after 798 seconds. |
| 21:11:47 | rb-bluebot | Build Server message: Revision 3664d382e0 result: All green |
| 21:11:54 | rb-bluebot | Build Server message: New build round started. Revision 42b63daebb, 345 builds, 8 clients. |
| 21:11:54 | rb-bluebot | ipod6g: Swap STORAGE_NEEDS_BOUNCE_BUFFER for STORAGE_WANTS_ALIGN by Solomon Peachy |
| 21:23:58 | rb-bluebot | Build Server message: Build round completed after 724 seconds. |
| 21:24:00 | rb-bluebot | Build Server message: Revision 42b63daebb result: All green |
| 21:57:10 | | Quit Moriar (Quit: Leaving.) |
| 22:00 |
| 22:26:48 | *** | Saving seen data "./dancer.seen" |
| 23:00 |
| 23:07:53 | | Quit massiveH (Quit: Leaving) |
icon identifies that the person is a core developer (has commit access).