| # 06:42:50 | y_morin | joins #crosstool-ng |
| # 07:04:19 | kos_tom | joins #crosstool-ng |
| # 07:27:41 | mnt_real | quits : Remote host closed the connection |
| # 08:41:01 | y_morin | quits : Quit: BBL |
| # 08:46:26 | y_morin | joins #crosstool-ng |
| # 09:43:15 | kos_tom | quits : Ping timeout: 255 seconds |
| # 09:53:36 | ctngbot | quits : Ping timeout: 255 seconds |
| # 10:08:06 | ctngbot_ | joins #crosstool-ng |
| # 10:25:03 | tkil | joins #crosstool-ng |
| # 10:25:48 | tkil | hm. are there any hooks in ct-ng for building 'perf' for the target? |
| # 11:26:42 | y_morin | tkil: nope. AFAIK, that's purely an application issue, and thus does not belong to the toolchain. |
| # 11:28:41 | tkil | yeah. relies on a bunch of low-level libs, tho. |
| # 11:28:56 | y_morin | tkil: what libs? |
| # 11:29:05 | tkil | libelf and/or elfutils (libdw in particular) |
| # 11:29:23 | tkil | having fun ith that now, apparently it intentionally uses "mktemp" instead of "mkstemp", but it's also built with -Werror... |
| # 11:30:11 | tkil | and it's an ulrich drepper production, at least in part... with his "unique" flavor of email interactions. :( |
| # 11:30:28 | y_morin | tkil: crosstool-NG does build libelf for the target, but only internally for use by ltrace. |
| # 11:31:01 | tkil | yes, i saw that. tried to use it, not sure if it's taking or not, not sure if perf really just wants elfutils. |
| # 11:31:35 | y_morin | Really, I believe this is an application issue. Probably, we could offer libelf for target being available even if ltrace is not built, but I have doubts about the appropriateness of such config... |
| # 11:34:22 | tkil | fair enough. |
| # 11:35:49 | tkil | although i'd make at least a partial argument that the kernel and kernel utils are just as much a part of the platform as the libc, so it's not clear where the ct-ng-generated platform ends and the application begins. |
| # 11:37:11 | y_morin | tkil: I see, but the kernel used to build the toolchain is only providing headers. The perf you want is the one coming with the kernel you actually run on the target, which might be different from the kernel used to build the toolchain. |
| # 11:37:38 | tkil | i do understand that; i just wonder how common that is. |
| # 11:38:02 | tkil | why would you use a different kernel to build the toolchain than the exact version you're planning on running on it? |
| # 11:39:24 | y_morin | tkil: for example, toolchain is built and validated (which takes time) by a team, and devel is done by another team, which may want to test a newer (even older!) kernel. |
| # 11:40:15 | tkil | hm. ok. although i'm not sure you can call a toolchain validated if you're swapping the kernel out from under the libc... |
| # 11:41:19 | tkil | (and, obviously, you can have a toolchain on top of bare metal...) |
| # 11:41:33 | tkil | anyway. |
| # 11:43:17 | y_morin | tkil: not (directly) related. For example, glibc can be configured with headers from a version, but with support for older versions (see --with-kernel=...) |
| # 11:43:57 | y_morin | And you can use a newer kernel with an older glibc. If all you want is a newer driver, and not use newer features (eg. AIO), then there's no need to rebuild the toolchain. |
| # 11:45:02 | tkil | fair enough. |
| # 11:56:24 | kos_tom | quits : Quit: Leaving |
| # 11:57:16 | kos_tom | joins #crosstool-ng |
| # 12:23:03 | kos_tom | quits : Ping timeout: 276 seconds |
| # 12:35:22 | kos_tom | joins #crosstool-ng |
| # 13:29:30 | ctngbot | joins #crosstool-ng |
| # 14:17:56 | mnt_real | joins #crosstool-ng |
| # 14:55:15 | y_morin | quits : Quit: Back on-line in a few hours... |
| # 16:43:19 | y_morin | joins #crosstool-ng |
| # 17:20:35 | kos_tom | quits : Quit: Leaving |
| # 17:20:38 | linuxjacques_ | joins #crosstool-ng |
| # 17:27:38 | y_morin | is now known as: y_morin|away |
| # 17:40:17 | y_morin|away | is now known as: y_morin |
| # 17:50:04 | kos_tom | joins #crosstool-ng |
| # 18:22:03 | y_morin | is now known as: y_morin|away |
| # 18:30:56 | y_morin|away | is now known as: y_morin |
| # 18:45:53 | y_morin | is now known as: y_morin|away |
| # 19:45:31 | y_morin|away | is now known as: y_morin |
| # 20:37:48 | bhundven|afk | is now known as: bhundven |
| # 20:37:57 | bhundven | quits : Changing host |
| # 20:37:57 | bhundven | joins #crosstool-ng |
| # 21:49:26 | y_morin | quits : Quit: Night all! See ya! :-) |