2016-05-02T01:22:49 *** printk has quit IRC 2016-05-02T01:22:49 *** prk has quit IRC 2016-05-02T01:26:49 *** prk has joined #rtems 2016-05-02T01:26:50 *** printk has joined #rtems 2016-05-02T01:55:47 *** gigetoo has quit IRC 2016-05-02T01:56:33 *** gigetoo has joined #rtems 2016-05-02T02:11:23 *** prk has quit IRC 2016-05-02T02:16:46 *** prk has joined #rtems 2016-05-02T03:45:24 *** mkhoory-3 has joined #rtems 2016-05-02T05:24:13 *** mkhoory-3 has quit IRC 2016-05-02T05:26:34 *** mkhoory has joined #rtems 2016-05-02T06:06:21 *** monstr has joined #rtems 2016-05-02T06:09:26 *** mkhoory-2 has joined #rtems 2016-05-02T06:12:44 *** mkhoory has quit IRC 2016-05-02T10:05:59 [git] ddc12de by Sebastian Huber: [bsp/qoriq: Add DCFG to memory map] http://devel.rtems.org/changeset/ddc12de/rtems 2016-05-02T10:06:09 [git] 0000791 by Sebastian Huber: [bsp/qoriq: Add and use qoriq_reset_qman_and_bman()] http://devel.rtems.org/changeset/0000791/rtems 2016-05-02T10:06:10 *** sebhub has joined #rtems 2016-05-02T10:06:18 [git] 3e36a16 by Sebastian Huber: [posix: _POSIX_Message_queue_Create_support() Make _POSIX_Message_queue_Create_support() static since it is only used by mq_open().] http://devel.rtems.org/changeset/3e36a16/rtems 2016-05-02T10:06:27 [git] 49c293f by Sebastian Huber: [posix: POSIX_Message_queue_Control::process_shared Delete unused POSIX_Message_queue_Control::process_shared.] http://devel.rtems.org/changeset/49c293f/rtems 2016-05-02T10:06:36 [git] dd6c28e by Sebastian Huber: [posix: Delete POSIX_Message_queue_Control::named Delete unused POSIX_Message_queue_Control::named.] http://devel.rtems.org/changeset/dd6c28e/rtems 2016-05-02T10:06:46 [git] 7580995 by Sebastian Huber: [score: _CORE_message_queue_Close() Move lock acquire to caller of _CORE_message_queue_Close() to allow state checks during object close operations under lock protection. Ensures deletion safety on uni-processor configuration.] http://devel.rtems.org/changeset/7580995/rtems 2016-05-02T10:06:55 [git] b0eba5e by Sebastian Huber: [score: _CORE_message_queue_Seize() Move lock acquire to caller of _CORE_message_queue_Seize() to allow state checks during receive operations under lock protection.] http://devel.rtems.org/changeset/b0eba5e/rtems 2016-05-02T10:07:04 [git] 99112f7 by Sebastian Huber: [score: _CORE_message_queue_Submit() Move lock acquire to caller of _CORE_message_queue_Submit() to allow state checks during send operations under lock protection.] http://devel.rtems.org/changeset/99112f7/rtems 2016-05-02T10:07:13 [git] 1b7a6a3 by Sebastian Huber: [score: _CORE_message_queue_Set_message_priority() Remove _CORE_message_queue_Set_message_priority() and set the priority in _CORE_message_queue_Insert_message().] http://devel.rtems.org/changeset/1b7a6a3/rtems 2016-05-02T10:07:23 [git] 88e09b9 by Sebastian Huber: [score: _CORE_message_queue_Insert_message() Move common code into _CORE_message_queue_Insert_message().] http://devel.rtems.org/changeset/88e09b9/rtems 2016-05-02T10:07:32 [trac] #2555/enhancement (accepted) updated by Sebastian Huber (In [changeset:"7e66865e17d7a82add541056de13717793da002a/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2555 2016-05-02T10:07:32 [git] 7e66865 by Sebastian Huber: [score: Move message notification Move message notification to end of critical section and delegate the message queue release to the notification handler. It may do more complex notification actions out of the critical section. Update #2555.] http://devel.rtems.org/changeset/7e66865/rtems 2016-05-02T10:07:42 [trac] #2555/enhancement (accepted) updated by Sebastian Huber (In [changeset:"6741d30a310695f4fe4c8acccb86ccf7389567bd/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2555 2016-05-02T10:07:43 [git] 6741d30 by Sebastian Huber: [rtems: Avoid Giant lock for message queues Update #2555.] http://devel.rtems.org/changeset/6741d30/rtems 2016-05-02T10:07:53 [trac] #2555/enhancement (accepted) updated by Sebastian Huber (In [changeset:"c8982e5f6a4857444676165deab1e08dc91a6847/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2555 2016-05-02T10:07:53 [trac] #2702/defect (new) updated by Sebastian Huber (In [changeset:"c8982e5f6a4857444676165deab1e08dc91a6847/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2702 2016-05-02T10:07:54 [git] c8982e5 by Sebastian Huber: [posix: Simplify message queues The mq_open() function returns a descriptor to a POSIX message queue object identified by a name. This is similar to sem_open(). In contrast to the POSIX semaphore the POSIX message queues use a separate object for the descriptor. This extra object is superfluous, since the object identifier 2016-05-02T10:07:54 can be used directly for t...] http://devel.rtems.org/changeset/c8982e5/rtems 2016-05-02T10:08:03 [trac] #2555/enhancement (accepted) updated by Sebastian Huber (In [changeset:"f009ed086d3da813a2c92b9834c3b2d618894883/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2555 2016-05-02T10:08:03 [git] f009ed0 by Sebastian Huber: [rtems: Avoid Giant lock for semaphores Update #2555.] http://devel.rtems.org/changeset/f009ed0/rtems 2016-05-02T10:08:13 [trac] #2555/enhancement (accepted) updated by Sebastian Huber (In [changeset:"f4d541ccfe6a6e64a06cbf89bc7b03e045f62281/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2555 2016-05-02T10:08:13 [git] f4d541c by Sebastian Huber: [rtems: Avoid Giant lock in rtems_object_set_name() Update #2555.] http://devel.rtems.org/changeset/f4d541c/rtems 2016-05-02T10:08:22 [trac] #2555/enhancement (accepted) updated by Sebastian Huber (In [changeset:"77726405fab31a04834e95968d1942fc81d0644d/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2555 2016-05-02T10:08:23 [git] 7772640 by Sebastian Huber: [score: _Objects_Get_name_as_string() Avoid Giant lock in _Objects_Get_name_as_string(). Update #2555.] http://devel.rtems.org/changeset/7772640/rtems 2016-05-02T10:08:32 [git] 3888771 by Sebastian Huber: [score: Streamline set time of day functions Rename _TOD_Set() into _TOD_Set_with_timespec(). Rename _TOD_Set_with_timestamp() into _TOD_Set(). This is now in line with _TOD_Get() and _TOD_Get_as_timespec(). The timestamp is the canonical format.] http://devel.rtems.org/changeset/3888771/rtems 2016-05-02T10:08:41 [trac] #2555/enhancement (accepted) updated by Sebastian Huber (In [changeset:"1ef8e4a8e93a848e2a68f37e029039300f1c936b/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2555 2016-05-02T10:08:42 [trac] #2630/enhancement (new) updated by Sebastian Huber (In [changeset:"1ef8e4a8e93a848e2a68f37e029039300f1c936b/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2630 2016-05-02T10:08:42 [git] 1ef8e4a by Sebastian Huber: [score: Avoid Giant lock for set time of day Update #2555. Update #2630.] http://devel.rtems.org/changeset/1ef8e4a/rtems 2016-05-02T10:08:51 [trac] #2124/defect (closed) updated by Sebastian Huber (In [changeset:"500a8e9c62dca9f62611ecca64857dadb2bc0557/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2124 2016-05-02T10:08:52 [git] 500a8e9 by Sebastian Huber: [score: Delete __RTEMS_STRICT_ORDER_MUTEX__ Remove support for strict order mutexes. Close #2124.] http://devel.rtems.org/changeset/500a8e9/rtems 2016-05-02T10:09:01 [git] d5cc9fd by Sebastian Huber: [score: __RTEMS_DO_NOT_INLINE_CORE_MUTEX_SEIZE__ Delete __RTEMS_DO_NOT_INLINE_CORE_MUTEX_SEIZE__ as a preparation to restructure the CORE mutex variants and reduce the branch complexity.] http://devel.rtems.org/changeset/d5cc9fd/rtems 2016-05-02T10:09:11 [trac] #2555/enhancement (accepted) updated by Sebastian Huber (In [changeset:"259d885168b548b5ebea5067f112a0d8b94d167f/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2555 2016-05-02T10:09:11 [git] 259d885 by Sebastian Huber: [posix: Remove superfluous thread dispatch disable The _Thread_queue_Enqueue_critical() already deals with thread dispatching. Update #2555.] http://devel.rtems.org/changeset/259d885/rtems 2016-05-02T10:09:20 [trac] #2690/defect (new) updated by Sebastian Huber (In [changeset:"66374dfc0c42066fac5f9a62475029711baac735/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2690 2016-05-02T10:09:20 [trac] #2555/enhancement (accepted) updated by Sebastian Huber (In [changeset:"66374dfc0c42066fac5f9a62475029711baac735/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2555 2016-05-02T10:09:21 [git] 66374df by Sebastian Huber: [posix: Avoid Giant lock in _POSIX_signals_Send() Update #2555. Update #2690.] http://devel.rtems.org/changeset/66374df/rtems 2016-05-02T10:09:30 [trac] #2510/enhancement (new) updated by Sebastian Huber (In [changeset:"981eed21761cd0036f0ac61042adea09d8a595a2/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2510 2016-05-02T10:09:31 [git] 981eed2 by Sebastian Huber: [score: Add dummy Strong APA scheduler Start with a copy of the Priority SMP scheduler implementation. Update #2510.] http://devel.rtems.org/changeset/981eed2/rtems 2016-05-02T10:37:22 *** shelf has quit IRC 2016-05-02T10:37:26 *** shelf has joined #rtems 2016-05-02T10:37:28 *** shelf has joined #rtems 2016-05-02T11:47:42 [git] b1b8bd2 by Sebastian Huber: [mpci: Fix _CORE_barrier_Surrender() Bug introduced by f27383a518836881b7b9b88e88d2e31d5b23d048.] http://devel.rtems.org/changeset/b1b8bd2/rtems 2016-05-02T11:47:42 [git] c506158 by Sebastian Huber: [mpci: Make _*_MP_Send_response_packet() static] http://devel.rtems.org/changeset/c506158/rtems 2016-05-02T11:47:42 [git] e38a92b by Sebastian Huber: [mpci: Make _*_MP_Get_packet() static] http://devel.rtems.org/changeset/e38a92b/rtems 2016-05-02T11:47:42 [git] 0a68d8e by Sebastian Huber: [score: Optimize _Objects_Get_local() Make the interrupt lock context the second parameter to avoid register moves.] http://devel.rtems.org/changeset/0a68d8e/rtems 2016-05-02T13:52:46 *** jkridner has joined #rtems 2016-05-02T14:46:54 *** printk has left #rtems 2016-05-02T15:25:23 *** monstr has quit IRC 2016-05-02T15:55:02 *** gigetoo has quit IRC 2016-05-02T15:56:59 *** gigetoo has joined #rtems 2016-05-02T18:07:51 *** gedare has joined #rtems 2016-05-02T18:07:51 *** ChanServ sets mode: +o gedare 2016-05-02T20:23:23 *** edwardk_ has joined #rtems 2016-05-02T20:28:39 *** darnir- has joined #rtems 2016-05-02T20:29:05 *** edwardk has quit IRC 2016-05-02T20:29:09 *** darnir has quit IRC 2016-05-02T20:29:16 *** darnir- is now known as darnir 2016-05-02T20:31:44 *** edwardk_ is now known as edwardk 2016-05-02T22:37:37 *** jkridner has quit IRC 2016-05-03T02:08:32 *** jkridner has joined #rtems 2016-05-03T02:14:53 *** jkridner has quit IRC 2016-05-03T02:17:56 *** ita has quit IRC 2016-05-03T02:18:11 *** ita has joined #rtems 2016-05-03T05:18:51 *** monstr has joined #rtems 2016-05-03T05:23:37 *** monstr has quit IRC 2016-05-03T05:35:18 *** monstr has joined #rtems 2016-05-03T05:41:49 *** jkridner has joined #rtems 2016-05-03T05:47:01 [trac] GSoC/2016 edited by hwj1593 (empty comment) http://devel.rtems.org/wiki/GSoC/2016 2016-05-03T05:47:49 *** jkridner has quit IRC 2016-05-03T05:58:10 sebhub: hi 2016-05-03T05:58:41 I see you added this code last year 'f91fbbf4 (Sebastian Huber 2015-09-30 15:30:09 +0200 102) extern rtems_i8259_masks i8259s_super_imr;' 2016-05-03T05:58:57 The IMR handling in the PC BSP with the atpic is rather broken 2016-05-03T07:58:16 *** shelf has quit IRC 2016-05-03T07:58:24 *** edwardk has quit IRC 2016-05-03T08:01:04 *** shelf has joined #rtems 2016-05-03T08:05:37 *** mkhoory-2 is now known as mkhoory 2016-05-03T08:08:41 *** edwardk has joined #rtems 2016-05-03T08:09:21 I added this to support the interrupt server necessary for the libbsd 2016-05-03T08:10:02 it worked at least on the pci network interface cards we used to test 2016-05-03T08:11:06 I really don't know how this works in detail 2016-05-03T08:11:49 did this change broke something that worked before? 2016-05-03T08:14:32 There is a long standing issue of spurious 7 interrupts which are basically a hardware race condition between the request line and PIC. I am seeing them. 2016-05-03T08:15:07 did I introduce this issue? 2016-05-03T08:15:10 I see the generic interrupt server does an enable, if the disable implictic in the PIC handling code. 2016-05-03T08:15:12 No 2016-05-03T08:15:34 It is related to changes in PC hardware and it moved to PCI 2016-05-03T08:16:06 i know next to nothing about the x86 architecture 2016-05-03T08:16:07 The ATPIC (legacy) and the PIC with edge for ISA bus ints, eg timer, keyboard, and level for PCI 2016-05-03T08:16:38 I have been over all the related FreeBSD code. It is to have such a good reference. 2016-05-03T08:17:21 yes, I looked at it and decided that I don't want to know the details 2016-05-03T08:17:29 My reading of your change is unless there is a disable call the IRQ is unmasked after the interrupt returns. 2016-05-03T08:17:44 And PCI ints are configured as level by the BIOS 2016-05-03T08:17:49 in legacy mode 2016-05-03T08:18:18 I have removed all the PIC programming code in the asm file and moved it all to C. 2016-05-03T08:18:32 The asm file is much simpler. 2016-05-03T08:19:07 for libbsd it is essential that you can disable the interrupt in the interrupt server interrupt handler routine and then later service the interrupt in the interrupt server thread context 2016-05-03T08:19:20 I can now manage the PIC and edges so if the IRQ is level, ie PCI, I keep it masked and wait for the server to run and then enable 2016-05-03T08:19:38 Cool, that is what I wanted to know 2016-05-03T08:19:46 This is not happening 2016-05-03T08:20:21 I am not sure why the code is not locking up in the interrupt but that is for another day. 2016-05-03T08:20:33 on BSD the interrupt service routines can block on mutexes, etc. so we need a thread context 2016-05-03T08:21:32 the freebsd kernel is a nice piece of software 2016-05-03T08:22:00 The FreeBSD interrupt model is nice, it is pre-thread, post-thread and post-filter handles the arch provides. For the ATPIC (non-APIC) it is EOI and disable for level for the pre-thread, EOI for the post-filter is not threads queued, and enable for the post-thread. 2016-05-03T08:23:03 yes, the problem is that interrupt handling is a part of the BSP 2016-05-03T08:23:03 Yes, that is all working nicely. It is just spurious ints and then we went to the default handler and then issued and EOI to the PIC which is wrong, it could have acked a prending int. 2016-05-03T08:23:42 I think the idea of moving to indirect calls for the pre/post is a nice idea. 2016-05-03T08:23:49 APIC will be interesting. 2016-05-03T08:24:09 Do we have an API for wiring IRQs to CPUs? 2016-05-03T08:24:35 an api yes (in a ticket or wiki page), but no implementation 2016-05-03T08:24:41 OK 2016-05-03T08:24:45 Nice anyway. 2016-05-03T08:25:21 With the nexu bus set up do we just add to the list for a BSP? 2016-05-03T08:25:29 for the PC this may become large 2016-05-03T08:26:05 it talked about this with joel before 2016-05-03T08:26:13 and? 2016-05-03T08:26:20 for the x86 we should use the FreeBSD nexus device 2016-05-03T08:26:33 File? 2016-05-03T08:26:50 it don't know off hand 2016-05-03T08:26:54 OK 2016-05-03T08:27:04 How does this help? 2016-05-03T08:27:22 I have the nexus bus working at the moment 2016-05-03T08:27:29 sys/x86/x86/nexus.c 2016-05-03T08:27:40 thanks 2016-05-03T08:28:28 This web site is nice http://fxr.watson.org/fxr/source/x86/x86/nexus.c, Amar pointed it out today 2016-05-03T08:28:59 Maybe we can look at this x86_64 and APCI and ACPI 2016-05-03T08:29:13 When those happen the switch can happen 2016-05-03T08:30:10 The PCI probe code we have is also broken, it locks up on certain devices in specific configurations on PCIe buses. 2016-05-03T08:30:30 the pc bsp is state of the art 1996 2016-05-03T08:30:36 Some chips lock the bus up 2016-05-03T08:31:04 Yeah, I have my 90's disco music playing while I hack on it 2016-05-03T08:31:44 We need to update it as some boards have dropped legacy support. 2016-05-03T08:31:57 Minnow is only UEFI boot 2016-05-03T08:34:13 Can I call bsp_interrupt_lock() with in an interrupt context? 2016-05-03T08:35:20 I need a call in irq-generic that lets me know if a handler is empty or not that I can call while in a interrupt context. 2016-05-03T08:35:57 http://fxr.watson.org/fxr/source/x86/isa/atpic.c#L545 2016-05-03T08:36:08 Is the FreeBSD piece of code I need to implement. 2016-05-03T08:38:20 you can test if the handler is bsp_interrupt_handler_default 2016-05-03T08:40:07 How? 2016-05-03T08:40:41 index = bsp_interrupt_handler_index(vector); 2016-05-03T08:40:47 head = &bsp_interrupt_handler_table [index]; 2016-05-03T08:40:51 empty = bsp_interrupt_is_empty_handler_entry(head); 2016-05-03T08:40:53 ? 2016-05-03T08:42:08 Do I need to call bsp_interrupt_lock() ? 2016-05-03T08:44:35 this is a mutex, so don't do this in interrupt context 2016-05-03T08:45:07 How manage locking or can I not do this? 2016-05-03T08:45:31 see irq-server.c, its a bit difficult 2016-05-03T08:45:56 and depends on the contxt 2016-05-03T08:47:17 Should I change irq-generic to also have an interrupt lock for this case? 2016-05-03T08:47:33 no, I tried hard to avoid this 2016-05-03T08:47:53 Just in the table updates and checks 2016-05-03T08:48:12 table updates are protected by the mutex 2016-05-03T08:48:26 Hmm ok. I will look some more. 2016-05-03T08:48:28 or the bsp_interrupt_vector_disable() 2016-05-03T08:48:57 We may need to add a BSP to allow it to have some low level bits to help here 2016-05-03T08:49:08 the code assumes that at most one interrupt service routine serves an interrupt at a time 2016-05-03T08:49:09 a BSP hook when installing 2016-05-03T08:49:42 This is true at the point of the test however after the mask is lowered to allow nesting 2016-05-03T08:51:34 Thanks for this, I will take another look tomorrow and see how I can do this. 2016-05-03T08:51:41 cu 2016-05-03T08:53:07 realtek up on a i7 with libbsd. 2016-05-03T10:04:47 *** jkridner has joined #rtems 2016-05-03T11:56:40 *** jkridner has quit IRC 2016-05-03T12:47:10 *** prk has quit IRC 2016-05-03T12:50:58 *** prk has joined #rtems 2016-05-03T14:08:45 [trac] GSoC/2016/RTEMStraceimprovements created http://devel.rtems.org/wiki/GSoC/2016/RTEMStraceimprovements 2016-05-03T14:09:36 [trac] GSoC/2016/RTEMStraceimprovements edited by vivekkukreja (empty comment) http://devel.rtems.org/wiki/GSoC/2016/RTEMStraceimprovements 2016-05-03T14:13:33 [trac] GSoC/2016 edited by vivekkukreja (empty comment) http://devel.rtems.org/wiki/GSoC/2016 2016-05-03T15:01:02 *** prk has quit IRC 2016-05-03T15:34:00 *** printk has joined #rtems 2016-05-03T15:47:53 github is not safe, my account is hacked today. 2016-05-03T17:04:08 *** gedare_ has joined #rtems 2016-05-03T17:25:07 *** sirfranklin has quit IRC 2016-05-03T17:27:49 *** sirfranklin has joined #rtems 2016-05-03T18:22:24 *** gedare_ has quit IRC 2016-05-03T18:38:14 *** gedare_ has joined #rtems 2016-05-03T18:42:58 *** gedare_ has quit IRC 2016-05-03T18:52:07 *** monstr has quit IRC 2016-05-03T19:37:43 [trac] GSoC/2016/Rock on RTEMS edited by Sambeet (empty comment) http://devel.rtems.org/wiki/GSoC/2016/Rock on RTEMS 2016-05-03T19:42:23 *** edwardk has quit IRC 2016-05-03T19:42:31 *** verm__ has quit IRC 2016-05-03T19:42:34 *** indy has quit IRC 2016-05-03T19:42:34 *** lcpfnvcy has quit IRC 2016-05-03T19:42:52 *** Boquete has quit IRC 2016-05-03T19:51:31 *** Boquete has joined #rtems 2016-05-03T19:51:32 *** edwardk has joined #rtems 2016-05-03T19:51:32 *** verm__ has joined #rtems 2016-05-03T19:51:32 *** indy has joined #rtems 2016-05-03T19:51:32 *** lcpfnvcy has joined #rtems 2016-05-03T21:13:37 *** hesham has quit IRC 2016-05-03T21:47:50 *** hesham has joined #rtems 2016-05-03T21:51:04 *** kappa1 has joined #rtems 2016-05-03T21:51:21 hi! 2016-05-03T21:51:30 I'm interested in participating in the ESA summer of code 2016-05-03T21:51:56 and this project caught my attention 2016-05-03T21:58:54 *** kappa1 has quit IRC 2016-05-04T04:44:54 *** mkhoory has quit IRC 2016-05-04T04:47:11 *** mkhoory has joined #rtems 2016-05-04T05:08:27 *** monstr has joined #rtems 2016-05-04T05:15:08 *** printk has quit IRC 2016-05-04T05:15:36 *** printk has joined #rtems 2016-05-04T05:25:27 [git] e4fd35d by Sebastian Huber: [confdefs.h: Fix named object size estimate Account for the terminating null character. Use _POSIX_PATH_MAX instead of NAME_MAX according to _POSIX_Semaphore_Manager_initialization() and _POSIX_Message_queue_Manager_initialization().] http://devel.rtems.org/changeset/e4fd35d/rtems 2016-05-04T05:25:28 [git] b1b6f3b by Sebastian Huber: [confdefs.h: Fix message queue size estimate Account for maximum message size alignment. Simplify _CORE_message_queue_Initialize().] http://devel.rtems.org/changeset/b1b6f3b/rtems 2016-05-04T05:25:28 [git] 159b637 by Sebastian Huber: [bsp/mvme5500: Use thread local variable Use thread local variable instead of task variable. Update #2494.] http://devel.rtems.org/changeset/159b637/rtems 2016-05-04T05:25:28 [git] 1d40d81 by Sebastian Huber: [rtems: Remove task variables Update #2494. Update #2555.] http://devel.rtems.org/changeset/1d40d81/rtems 2016-05-04T05:25:28 [git] 0ab4fb7 by Sebastian Huber: [score: Fix _Objects_Id_to_name() Avoid out-of-bounds array access in case the object class is invalid.] http://devel.rtems.org/changeset/0ab4fb7/rtems 2016-05-04T05:25:28 [git] 469dc47 by Sebastian Huber: [score: Simplify _Objects_Id_to_name()] http://devel.rtems.org/changeset/469dc47/rtems 2016-05-04T05:25:28 [git] 876dde7 by Sebastian Huber: [score: Make _Objects_Information_table const The _Objects_Information_table is statically initialized. So, we can make it read-only.] http://devel.rtems.org/changeset/876dde7/rtems 2016-05-04T05:25:29 [git] 8c77150 by Sebastian Huber: [score: Simplify _Thread_Get_objects_information()] http://devel.rtems.org/changeset/8c77150/rtems 2016-05-04T05:25:29 [git] 822ec58 by Sebastian Huber: [score: Add _Thread_MP_Is_remote()] http://devel.rtems.org/changeset/822ec58/rtems 2016-05-04T05:25:30 [git] 49a134d by Sebastian Huber: [rtems: Return status in _Event_Surrender()] http://devel.rtems.org/changeset/49a134d/rtems 2016-05-04T05:25:30 [git] a0dee7c by Sebastian Huber: [score: Simplify _Thread_Get_interrupt_disable() Remove the object location parameter.] http://devel.rtems.org/changeset/a0dee7c/rtems 2016-05-04T05:27:15 [trac] #2306/defect (closed) updated by sebastian.huber ([159b63701539f3e584668a4ace13c6109d3b364b/rtems]) http://devel.rtems.org/ticket/2306 2016-05-04T05:28:19 [trac] #2494/defect (new) updated by sebastian.huber ([1d40d81b4b8dd50e4162b0b79b60d3312d2744e5/rtems] Documentation update is missing.) http://devel.rtems.org/ticket/2494 2016-05-04T05:29:06 [trac] #2555/enhancement (accepted) updated by sebastian.huber ([1d40d81b4b8dd50e4162b0b79b60d3312d2744e5/rtems]) http://devel.rtems.org/ticket/2555 2016-05-04T05:31:18 sebhub: hi 2016-05-04T05:31:22 hi chris 2016-05-04T05:31:28 sebhub: the LICENSE file is in rtems-docs.git 2016-05-04T05:32:46 I am wondering about adding to each source file 'SPDX-License-Identifier: CC-BY-SA-4.0' 2016-05-04T05:34:44 very nice, thanks 2016-05-04T05:34:56 :) 2016-05-04T05:35:38 so, no excuse for not adding documentation anymore 2016-05-04T05:36:15 Please check the tickets for doco I have not finished converting over. 2016-05-04T05:36:15 yes, I think a license identifier per file would be good 2016-05-04T05:36:34 Great. I will add it. We do not want a full license per file. 2016-05-04T05:37:06 yes, that is annoying 2016-05-04T05:37:07 Amar will move the figures to a common area. This was not done from the start but it is a good idea. 2016-05-04T05:37:44 we have some budget for documentation after the SMP implementation phase 2016-05-04T05:38:15 Great. May be we should discuss the structure a little more. It is still fragemented. 2016-05-04T05:38:22 when will the docs repo move to git.rtems.org? 2016-05-04T05:38:35 It is. 2016-05-04T05:38:37 and way too much copy and paste 2016-05-04T05:38:41 Yes 2016-05-04T05:38:54 its not listed here: https://git.rtems.org/ 2016-05-04T05:39:00 I will get the cgit, and posts added. 2016-05-04T05:39:40 who will remove the docs from the main repository? 2016-05-04T05:40:00 Oh we need to branch the repo 2016-05-04T05:40:27 Arrr, I need to finish converting the doco first or 4.11 will be left behind 2016-05-04T05:40:45 See this before "Limiting %s from %d to %d packets/sec" ? 2016-05-04T05:40:54 yes 2016-05-04T05:41:10 I will fix the log output this week or next week 2016-05-04T05:41:12 How can you stop the limiting? 2016-05-04T05:41:22 see freebsd documentation ;-) 2016-05-04T05:41:26 haha, nice 2016-05-04T05:41:44 I guess you need a consumer that is fast enough 2016-05-04T05:43:15 ping 2016-05-04T05:43:24 flood ping on a gige link 2016-05-04T05:43:54 I meant if you have someone that consumes the food packages fast enough, then there should be not rate limit 2016-05-04T05:44:01 Yes 2016-05-04T05:44:24 if you have a high packet drop rate, then its reasonable to impose a rate limit 2016-05-04T05:44:45 Might fix the debug print to get real formatting. 2016-05-04T05:45:01 yes, i will work on this in the next days 2016-05-04T05:45:13 you are not the first one complaining 2016-05-04T05:45:28 I am not complaining, I might fix soon. 2016-05-04T05:45:49 did you start to work on this? 2016-05-04T05:45:57 Not yet 2016-05-04T05:46:09 Still deep in the ATPIC on the PC. 2016-05-04T05:46:18 Headed back to 1976 doco on the PIC 2016-05-04T05:46:22 my plan is to import subr_prf.c and use it for log(), vlog(), syslog() and vsyslog() 2016-05-04T05:46:33 ok, please don't work on it 2016-05-04T05:46:36 Oh ok. I was not going to do that. 2016-05-04T05:46:41 I can wait 2016-05-04T05:47:43 My current changes are small and mostly PC related set up related 2016-05-04T05:48:31 OK to add the tcpdump command to testsuite/include/rtems/bsd/test/default-network-init.h? 2016-05-04T05:49:04 yes, of course 2016-05-04T05:50:52 [trac] #2703/defect (v:4.10) created by sebastian.huber (MPCI is broken on SMP configurations) http://devel.rtems.org/ticket/2703 2016-05-04T05:53:29 Do we have sysctl, ie '# sysctl net.inet.icmp.icmplim=0'? 2016-05-04T05:54:16 yes, didn't it work? 2016-05-04T05:55:28 I did not try, was look at how to port it, silly me :) 2016-05-04T05:55:38 [trac] #2555/enhancement (accepted) updated by Sebastian Huber (In [changeset:"b30ab250f0967920d6c08eae776fb82e717da182/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2555 2016-05-04T05:55:38 [trac] #2703/defect (new) updated by Sebastian Huber (In [changeset:"b30ab250f0967920d6c08eae776fb82e717da182/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2703 2016-05-04T05:55:39 [git] b30ab25 by Sebastian Huber: [mpci: Avoid Giant lock in _MPCI_Process_response() Update #2555. Update #2703.] http://devel.rtems.org/changeset/b30ab25/rtems 2016-05-04T05:56:19 the sysctl stuff is used by various configuration commands 2016-05-04T05:56:31 But a cli command? 2016-05-04T05:56:50 what is a cli command? 2016-05-04T05:57:00 command line interface 2016-05-04T05:57:19 "/sbin/sysctl" 2016-05-04T05:57:48 it would be nice to have it 2016-05-04T05:58:02 I will take a look, should be easy compared to tcpdump 2016-05-04T06:04:04 [git] 7b8f336 by Chris Johns: [Remove debug output.] http://devel.rtems.org/changeset/7b8f336/rtems-libbsd 2016-05-04T06:04:04 [git] b8d1fe0 by Chris Johns: [pc386: Add RealTek network driver to the nexus bus.] http://devel.rtems.org/changeset/b8d1fe0/rtems-libbsd 2016-05-04T06:04:04 [git] f3f816c by Chris Johns: [Add nexus devices for default-init.] http://devel.rtems.org/changeset/f3f816c/rtems-libbsd 2016-05-04T06:04:04 [git] f478e3c by Chris Johns: [Add sys/rpc so it is installed.] http://devel.rtems.org/changeset/f478e3c/rtems-libbsd 2016-05-04T06:04:04 [git] 1e8c93c by Chris Johns: [Add some doco on reverse and forward movement of the code.] http://devel.rtems.org/changeset/1e8c93c/rtems-libbsd 2016-05-04T06:04:04 [git] a1a2e10 by Chris Johns: [Add tcpdump to the test shell commands.] http://devel.rtems.org/changeset/a1a2e10/rtems-libbsd 2016-05-04T06:04:04 [git] 4b127e7 by Chris Johns: [Add RealTek Gig PHY.] http://devel.rtems.org/changeset/4b127e7/rtems-libbsd 2016-05-04T06:05:08 *** jkridner has joined #rtems 2016-05-04T06:24:20 [git] 5338089 by Chris Johns: [Fix coding to be CamelCase.] http://devel.rtems.org/changeset/5338089/rtems-libbsd 2016-05-04T07:04:41 Hmm 'net.inet6.ip6.kame_version: FreeBSD', should we change this? 2016-05-04T07:05:10 what is this? 2016-05-04T07:05:15 sysctl :) 2016-05-04T07:05:50 I have the command working 2016-05-04T07:05:56 Also 'kern.hostuuid: 00000000-0000-0000-0000-00000000000' 2016-05-04T07:07:02 Crashes in the net.bpf.buf something 2016-05-04T07:08:16 not tested == not working 2016-05-04T07:08:29 The command? 2016-05-04T07:08:46 the sysctls 2016-05-04T07:08:55 some execute handlers to get the values 2016-05-04T07:09:15 With the command we can think about tests via scripts 2016-05-04T07:09:27 dies in sysctl_netisr_work 2016-05-04T07:09:45 freebsd/sys/net/netisr.c:1424 2016-05-04T07:09:54 the nwsp is NULL 2016-05-04T07:10:17 DPCPU_ID_PTR ? 2016-05-04T07:16:20 the per cpu stuff is more or less disabled 2016-05-04T07:16:49 I see. I am just seeing how to get nwsp from nws via rtems_bsd_nws 2016-05-04T07:17:08 Not sure if this is &rtems_bsd_nws 2016-05-04T07:17:29 It is. 2016-05-04T07:29:30 Heap corruption now. 2016-05-04T08:53:01 sebhub: I just noticed I push a change with white space to libbsd. Grr 2016-05-04T08:53:28 too much white stuff on your desk? 2016-05-04T08:53:54 Hmm no white stuff on my desk 2016-05-04T09:50:12 [git] e58b898 by Sebastian Huber: [telnetd: Use syslog() instead of printk()] http://devel.rtems.org/changeset/e58b898/rtems-libbsd 2016-05-04T09:50:12 [git] 27aedff by Sebastian Huber: [default-network-init.h: Fix warnings] http://devel.rtems.org/changeset/27aedff/rtems-libbsd 2016-05-04T09:54:07 *** dougk_ has quit IRC 2016-05-04T10:06:40 *** dougk_ has joined #rtems 2016-05-04T10:22:43 *** sirfranklin has quit IRC 2016-05-04T10:23:34 *** sirfranklin has joined #rtems 2016-05-04T13:20:58 *** kappa1 has joined #rtems 2016-05-04T13:34:56 *** sebhub has quit IRC 2016-05-04T15:14:32 *** kappa1 has quit IRC 2016-05-04T15:39:42 *** printk has quit IRC 2016-05-04T16:03:01 *** monstr has quit IRC 2016-05-04T16:12:18 *** gedare has quit IRC 2016-05-04T16:41:10 *** gedare has joined #rtems 2016-05-04T16:41:10 *** ChanServ sets mode: +o gedare 2016-05-04T16:55:12 *** gedare has quit IRC 2016-05-04T18:03:38 *** gedare has joined #rtems 2016-05-04T18:03:38 *** ChanServ sets mode: +o gedare 2016-05-04T18:05:49 *** jkridner has quit IRC 2016-05-04T19:16:26 *** jkridner has joined #rtems 2016-05-04T19:21:10 *** jkridner has quit IRC 2016-05-04T19:36:24 *** gedare has quit IRC 2016-05-04T23:15:13 *** kiwichris_ has quit IRC 2016-05-05T05:21:34 *** aj007 has joined #rtems 2016-05-05T05:59:27 *** monstr has joined #rtems 2016-05-05T06:38:04 *** aj007 has quit IRC 2016-05-05T13:48:17 *** printk has joined #rtems 2016-05-05T13:49:02 *** gedare_ has joined #rtems 2016-05-05T14:05:38 *** jkridner has joined #rtems 2016-05-05T15:01:26 *** printk has quit IRC 2016-05-05T15:10:58 *** jkridner has quit IRC 2016-05-05T15:37:19 *** printk has joined #rtems 2016-05-05T16:08:09 *** printk has quit IRC 2016-05-05T16:15:36 *** printk has joined #rtems 2016-05-05T16:16:05 *** gedare_ has quit IRC 2016-05-05T17:30:03 *** dipupo has joined #rtems 2016-05-05T17:31:15 *** jkridner has joined #rtems 2016-05-05T17:32:22 *** printk has quit IRC 2016-05-05T17:32:42 *** printk has joined #rtems 2016-05-05T17:38:05 *** jkridner has quit IRC 2016-05-05T17:50:15 *** gedare_ has joined #rtems 2016-05-05T18:19:52 *** _sambeet has joined #rtems 2016-05-05T18:44:42 *** _sambeet has quit IRC 2016-05-05T18:59:53 *** gedare_ has quit IRC 2016-05-05T19:21:37 *** gedare_ has joined #rtems 2016-05-05T19:25:05 *** monstr has quit IRC 2016-05-05T19:37:22 *** gedare_ has quit IRC 2016-05-05T19:45:44 *** dipupo_ has joined #rtems 2016-05-05T19:57:31 *** gedare_ has joined #rtems 2016-05-05T20:02:42 *** gedare_ has quit IRC 2016-05-05T20:19:14 *** gigetoo has quit IRC 2016-05-05T20:20:05 *** gigetoo has joined #rtems 2016-05-05T21:18:36 [trac] GSoC/2016 edited by Habeeb (empty comment) http://devel.rtems.org/wiki/GSoC/2016 2016-05-05T22:34:03 *** ambro718 has joined #rtems 2016-05-05T22:36:16 Hello, can someone please tell me if this is possible. Say I have two network interfaces. I want to process received UDP packets on one of the interfaces in hard real-time, with tight worst-case latency requirements. Traffic on the other interface must not cause degradation of performance on this interface with super important UDP traffic. 2016-05-05T22:37:16 I have already looked in the Linux kernel code and determined that its design with the use of softirqs does not support this requirement. 2016-05-05T22:38:14 (it does not prioritize based on interfaces, just processes RX packets in the order they were queued, something like that) 2016-05-05T23:38:36 *** ambro718 has quit IRC 2016-05-06T00:39:02 *** printk has quit IRC 2016-05-06T00:40:31 *** dougk_ has quit IRC 2016-05-06T01:05:24 *** kiwichris_ has joined #rtems 2016-05-06T02:07:13 *** jkridner has joined #rtems 2016-05-06T03:02:21 *** gedare has joined #rtems 2016-05-06T03:02:21 *** ChanServ sets mode: +o gedare 2016-05-06T03:12:49 *** gedare has quit IRC 2016-05-06T05:24:39 *** sebhub has joined #rtems 2016-05-06T05:25:57 sebhub: hi. Just sent you an email with a network-config.h error I am seeing. Why was the header changed? 2016-05-06T05:27:18 *** monstr has joined #rtems 2016-05-06T05:41:36 sebhub: should "bld(target = "testsuite/include/rtems/bsd/test/network-config.h"," in libbsd_waf.py be "bld(target = "testsuite/include/network-config.h","? This means your change works. I like '#include "network-config.h";. 2016-05-06T05:42:55 the #include "..." should search in the directory of this file 2016-05-06T05:43:17 Why not move the file to the top and then nothing else needs to change? It is generated. 2016-05-06T05:43:44 so, network-config.h and default-network-init.h should be in the same directory 2016-05-06T05:44:27 I don't like a #include "..." which falls back to a #include <...> 2016-05-06T05:44:37 ? 2016-05-06T05:44:53 I try to figure out why this file is in different locations on your and my machine 2016-05-06T05:45:03 Oh, the generated file in the source is bad and should happen 2016-05-06T05:45:12 should not happen 2016-05-06T05:45:42 do you use --net-test-config=? 2016-05-06T05:46:19 no 2016-05-06T05:46:31 how do you get the header? 2016-05-06T05:46:57 i don't know 2016-05-06T05:47:07 Is this a hangover from the makefile days? 2016-05-06T05:47:22 It put generated source in the source tree 2016-05-06T05:49:06 I have sysctl command working. I found the memory corruption bug. This line is wrong ... 2016-05-06T05:49:09 https://git.rtems.org/rtems-libbsd/tree/freebsd/sys/sys/smp.h#n90 2016-05-06T05:49:36 I have a patch for it waiting until we have this header file sorted out 2016-05-06T05:49:56 why is this line wrong? 2016-05-06T05:50:07 The maxid for a single cpu is 0. 2016-05-06T05:50:21 See the CPU_FOREACH 2016-05-06T05:50:27 just below 2016-05-06T05:50:27 oh, that is bad 2016-05-06T05:50:29 <= 2016-05-06T05:50:32 Fixed 2016-05-06T05:51:03 Had to add i386 debug register support to the gdb stub to find it so it was a good exercise :) 2016-05-06T05:51:48 I will push my changes I have 2016-05-06T05:52:13 [git] 7a79eec by Chris Johns: [The maxid is not the number of cpus.] http://devel.rtems.org/changeset/7a79eec/rtems-libbsd 2016-05-06T05:52:13 [git] c60deba by Chris Johns: [Fix sysctl support.] http://devel.rtems.org/changeset/c60deba/rtems-libbsd 2016-05-06T05:52:13 [git] 83eb07c by Chris Johns: [Add sbin/sysctl] http://devel.rtems.org/changeset/83eb07c/rtems-libbsd 2016-05-06T05:52:13 [git] a51d0fe by Chris Johns: [Move network-config.h to the top of the testsuite include path.] http://devel.rtems.org/changeset/a51d0fe/rtems-libbsd 2016-05-06T05:52:48 Oops I pushed the network-config. I will fix it what we want once we figure it out. 2016-05-06T05:53:11 sysctl command is very nice 2016-05-06T05:54:20 I think we need something like sysctl.conf for us. I am not sure how we do it yet. If we have it things like this could be removed .. https://git.rtems.org/rtems-libbsd/tree/freebsd/sys/kern/kern_mbuf.c#n137 2016-05-06T05:54:25 the mp_maxid is also wrong in uma_core.c, I will fix this 2016-05-06T05:54:33 OK and thanks 2016-05-06T05:54:58 We should avoid adding special hacks like the mbuf mem size for RTEMS and use sysctl 2016-05-06T05:55:27 Joel as a NIC which needs 64M and not 8M and this should be set using sysctl 2016-05-06T05:55:42 no, its more difficult than that 2016-05-06T05:55:52 It is? 2016-05-06T05:56:27 yes, see rtems_bsd_get_allocator_domain_size() and how UMA gets its memory 2016-05-06T05:57:56 I saw that, what is the issue? 2016-05-06T05:59:26 This approach will not scale well. It breaks on existing cards 2016-05-06T05:59:47 see rtems-bsd-page.h and uma_int.h (rtems_bsd_page_get_object()) 2016-05-06T06:00:06 no, in this case your configuration is wrong 2016-05-06T06:00:23 In what way? 2016-05-06T06:00:23 you can provide your own implementation of rtems_bsd_get_allocator_domain_size() 2016-05-06T06:00:36 Tests do not run 2016-05-06T06:01:16 then the configuration needs to know how much memory it needs 2016-05-06T06:02:03 the UMA is absolutely performance critical 2016-05-06T06:02:54 Do we change the value and commit it? It would work for this box then 2016-05-06T06:03:16 which value? 2016-05-06T06:03:26 size = 8 * 1024 * 1024 2016-05-06T06:03:31 The 8M 2016-05-06T06:03:48 then your board works, other no longer since they don't have that much ram 2016-05-06T06:04:12 we had the same problem with the old network stack 2016-05-06T06:04:40 Yes I understand that and why I am saying the approach is wrong. I see no reason to have these presets done this way. 2016-05-06T06:04:59 I have not looked at the PAGE value but I will. 2016-05-06T06:05:24 FreeBSD uses sysctl and we should not invent new things and use what is done and documented. 2016-05-06T06:05:49 For example the BSP could provide a set of sysctl values to set 2016-05-06T06:06:03 see man sysctl.conf 2016-05-06T06:06:18 yes, I am not against this, but the UMA should stay at least as efficient as it is today even if it is not optimal configurable 2016-05-06T06:06:28 Yes of course. 2016-05-06T06:06:51 the network stack configuration is far away from being perfect 2016-05-06T06:06:52 In terms of the mbuf size the code becomes standard FreeBSD code 2016-05-06T06:07:07 We should use sysctl, it is how FreeBSD does it. 2016-05-06T06:07:25 yes, the rule of thumb should be to be as close to freebsd as possible 2016-05-06T06:07:28 We just need some sugar somewhere to make it easy for users. 2016-05-06T06:08:18 I am really happy that you work on the libbsd 2016-05-06T06:08:33 Been wanting to get to this stuff doe a while. 2016-05-06T06:08:45 Tyr the sysctl command. We have around 290 variables. 2016-05-06T06:09:46 I have an interesting issue. A flood ping gives a result of ... 2016-05-06T06:09:54 867 packets transmitted, 597 packets received, 31.1% packet loss 2016-05-06T06:10:10 and netstat -i gives ... 2016-05-06T06:10:12 re0 1500 50:e5:49:51:6c:b7 887 0 0 613 0 0 2016-05-06T06:10:52 The output packet count matches the number. I am yet to figure why some got dropped. 2016-05-06T06:12:47 netstat -s shows nothing wrong. 2016-05-06T06:14:16 maybe they are dropped at hardware level 2016-05-06T06:14:37 The input count matches the ping sent count 2016-05-06T06:15:07 The 20 difference is about the same as the sent difference 2016-05-06T06:18:14 [trac] #2555/enhancement (accepted) updated by Sebastian Huber (In [changeset:"bb2ad039a7246eecde65592a5116c86d3dede34b/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2555 2016-05-06T06:18:14 [git] bb2ad03 by Sebastian Huber: [rtems: Avoid Giant lock for signals Update #2555.] http://devel.rtems.org/changeset/bb2ad03/rtems 2016-05-06T06:18:27 [trac] #2555/enhancement (accepted) updated by Sebastian Huber (In [changeset:"64051ec80b55c3945ace509fa275c21ab8c89ab1/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2555 2016-05-06T06:18:28 [git] 64051ec by Sebastian Huber: [posix: Avoid Giant lock in pthread_equal() Update #2555.] http://devel.rtems.org/changeset/64051ec/rtems 2016-05-06T06:18:40 [trac] #2555/enhancement (accepted) updated by Sebastian Huber (In [changeset:"f65f803a269784f6f05658054dd1709497ca5049/rtems"]: {{{ ...) http://devel.rtems.org/ticket/2555 2016-05-06T06:18:40 [git] f65f803 by Sebastian Huber: [score: Avoid Giant lock for CBS scheduler Update #2555.] http://devel.rtems.org/changeset/f65f803/rtems 2016-05-06T06:24:18 [ 1/849] testsuite/include/rtems/bsd/test/network-config.h: testsuite/include/rtems/bsd/test/network-config.h.in -> testsuite/include/rtems/bsd/test/network-config.h 2016-05-06T06:24:20 [ 2/849] rtemsbsd/rtems/rtems-kvm-symbols.c: rtemsbsd/rtems/generate_kvm_symbols -> rtemsbsd/rtems/rtems-kvm-symbols.c 2016-05-06T06:24:21 [ 3/849] c: rtemsbsd/rtems/rtems-kvm-symbols.c -> build/arm-rtems4.12-xilinx_zynq_a9_qemu/rtemsbsd/rtems/rtems-kvm-symbols.c.3.o 2016-05-06T06:24:45 This is how network-config.h is generated on my machine 2016-05-06T06:25:46 [ 1/849] testsuite/include/rtems/bsd/test/network-config.h: testsuite/include/rtems/bsd/test/network-config.h.in -> build/arm-rtems4.12-xilinx_zynq_a9_qemu/testsuite/include/rtems/bsd/test/network-config.h 2016-05-06T06:25:47 [ 2/849] rtemsbsd/rtems/rtems-kvm-symbols.c: rtemsbsd/rtems/generate_kvm_symbols -> rtemsbsd/rtems/rtems-kvm-symbols.c 2016-05-06T06:25:49 [ 3/849] c: rtemsbsd/rtems/rtems-kvm-symbols.c -> build/arm-rtems4.12-xilinx_zynq_a9_qemu/rtemsbsd/rtems/rtems-kvm-symbols.c.3.o 2016-05-06T06:26:10 this is how it is genrated after I removed my network-config.h in the source directory 2016-05-06T06:26:19 looks like a bug in the build script 2016-05-06T06:26:24 or a feature 2016-05-06T06:31:42 waf 1.8.20 (14c787c991e1d9badd9d302bc9eaa060e9d4b03c) 2016-05-06T06:32:04 waf 1.7.0 (a7e69d6b81b04729804754c4d5214da063779a65) 2016-05-06T06:32:22 Maybe we should add waf to libbsd and remove this issue 2016-05-06T06:32:51 Or add a version check 2016-05-06T06:32:52 yes, but what happens on your machine after a "touch testsuite/include/rtems/bsd/test/network-config.h" ? 2016-05-06T06:32:55 or both 2016-05-06T06:33:29 It looks in the source tree, but let me check 2016-05-06T06:34:39 generated in the build tree 2016-05-06T06:34:44 so I have 2 now 2016-05-06T06:34:56 ok, it is waf version dependent 2016-05-06T06:35:05 Yes 2016-05-06T06:35:15 ita would know 2016-05-06T06:35:36 and the source version is included 2016-05-06T06:36:12 testsuite/ping01/test_main.c:40 is now wrong 2016-05-06T06:36:23 should I change this to "network-config.h" ? 2016-05-06T06:36:40 no, I revert your and my change 2016-05-06T06:36:49 OK 2016-05-06T06:37:03 I will leave this and get back to the spurious interrupt on the PC 2016-05-06T06:42:20 sebhub: make sure waf 1.8 is in use 2016-05-06T06:42:30 ita: thanks 2016-05-06T06:42:48 yes, I updated to this version 2016-05-06T06:43:08 I will add a check to the script to enforce this 2016-05-06T06:43:39 i am not sure what the problem is, but in general "waf -v" can display more diagnostics 2016-05-06T06:44:12 is a file generated in the source directory? 2016-05-06T06:44:31 No, it is in the build tree with 1.8.20 2016-05-06T06:45:30 I think the 1.8.20 behaviour is correct. 2016-05-06T06:45:37 if you ever end up with build files in the source directory, remove them and run "waf clean build" to be certain 2016-05-06T06:46:05 Yes. When we ran the makefile build in a parallel while we moved to waf this caused problems. 2016-05-06T06:46:17 The makefile support has gone 2016-05-06T06:47:27 ok, perfect 2016-05-06T06:48:55 we will soon (1-2 months) have waf 1.9 which will bring a few interesting features such as allowing rebuilds when users want to change generated files in the build directory (headers...) 2016-05-06T06:50:30 there will be no need to change the rtems scripts 2016-05-06T06:51:40 Great. 2016-05-06T06:51:53 (full list of changes -> https://github.com/waf-project/waf/blob/waf-1.9/ChangeLog) 2016-05-06T06:55:45 ita: What does "Make lazy visual studio detection the default" mean? 2016-05-06T06:58:15 kiwichris_: in waf 1.8 there is a --msvc_lazy_autodetect flag that enables faster configuration when several versions are present on the system (querying registry keys and running bat files can be slow) 2016-05-06T06:59:47 [git] 542c981 by Sebastian Huber: [ZONE(9): Fix mp_maxid definition] http://devel.rtems.org/changeset/542c981/rtems-libbsd 2016-05-06T06:59:47 [git] ffbac33 by Sebastian Huber: [Revert "Move network-config.h to the top of the testsuite include path." This reverts commit a51d0fea6880046300fcf466a767f88eab9a6f9a.] http://devel.rtems.org/changeset/ffbac33/rtems-libbsd 2016-05-06T06:59:47 [git] ee9991f by Sebastian Huber: [Partially revert 27aedffc1bd8e41f4e06a3326a5b95c54ca38b79] http://devel.rtems.org/changeset/ee9991f/rtems-libbsd 2016-05-06T06:59:47 [git] 0f5dd1c by Sebastian Huber: [Add script to create the kernel namespace header] http://devel.rtems.org/changeset/0f5dd1c/rtems-libbsd 2016-05-06T07:00:06 kiwichris_: this option is could not be made the default in 1.8 because the full list of compilers is added to a conf.env.MSVC_INSTALLED_VERSIONS 2016-05-06T07:00:33 ita: thanks 2016-05-06T07:02:02 sebhub: I needed 'sysctl net.inet.icmp.icmplim_output=0' and 'sysctl net.inet.icmp.icmplim=30000' and now I can flood ping 2016-05-06T07:02:25 The 30000 is a big number, I should try 0 2016-05-06T07:02:47 Yes 0 is the proper value to disable the rate limiting 2016-05-06T07:03:21 its nice that we can use the sysctl on the command line now 2016-05-06T07:04:29 what is really nice with waf is that you don't need the tools in your $PATH 2016-05-06T07:05:15 Too many things I like to list here :) 2016-05-06T07:12:11 sebhub: when should the new script be run? 2016-05-06T07:12:37 whenever you add a global symbol in the kernel space 2016-05-06T07:13:00 for example? 2016-05-06T07:13:01 *** monstr has quit IRC 2016-05-06T07:13:13 me adding new drivers? 2016-05-06T07:13:19 yes 2016-05-06T07:13:51 its a hack, the script output depends on the build configurations 2016-05-06T07:14:18 OK. Should libbsd.txt have something about this? 2016-05-06T07:15:00 This is much better "1990360 packets transmitted, 1990359 packets received, 0.0% packet loss" 2016-05-06T07:15:32 ok, I will add something like this, its difficult to find the right spot in libbsd.txt 2016-05-06T07:16:28 Any place will do. We need doco added to the docs repo and so this file will become developer focused 2016-05-06T07:27:24 *** monstr has joined #rtems 2016-05-06T07:29:15 sebhub: flood ping with 25000 byte packets with rx and tx both at 63K pps and 92MB/s on the rx and tx. The i7 is using 15% in IRQS and 23% in the stack. 2016-05-06T07:32:42 [git] 28c973a by Sebastian Huber: [Add "How to import code from FreeBSD"] http://devel.rtems.org/changeset/28c973a/rtems-libbsd 2016-05-06T07:35:58 *** ambro718 has joined #rtems 2016-05-06T07:52:17 with waf 1.8.20 libbsd.a and thus the tests are now linked each time 2016-05-06T07:52:38 with the waf 1.7 libbsd.a was linked each time, but not the tests 2016-05-06T07:53:23 Sorry, is this good or bad? 2016-05-06T07:53:44 good would be not not link at all 2016-05-06T07:54:06 I am lost. 2016-05-06T07:54:32 If you change the code the tests need to be linked 2016-05-06T07:54:41 a "waf build" links the libbsd.a and all tests even if I didn't touch anything 2016-05-06T07:54:53 That does not happen for me 2016-05-06T07:54:57 waf distclean 2016-05-06T07:55:01 waf configure .... 2016-05-06T07:55:15 *** ambro718 has quit IRC 2016-05-06T07:56:13 *** peerst_ has joined #rtems 2016-05-06T07:56:49 self:/scratch/git-rtems-libbsd (upstream) > waf distclean 2016-05-06T07:56:50 'distclean' finished successfully (0.235s) 2016-05-06T07:56:52 self:/scratch/git-rtems-libbsd (upstream) > waf configure --prefix=/build/rtems-4.12 --rtems-bsps=arm/xilinx_zynq_a9_qemu 2016-05-06T07:56:53 Setting top to : /scratch/git-rtems-libbsd 2016-05-06T07:56:55 Setting out to : /scratch/git-rtems-libbsd/build 2016-05-06T07:56:56 RTEMS Version : 4.12 2016-05-06T07:56:58 Architectures : arm-rtems4.12, bfin-rtems4.12, epiphany-rtems4.12, i386-rtems4.12, lm32-rtems4.12, m32c-rtems4.12, m68k-rtems4.12, microblaze-rtems4.12, mips-rtems4.12, moxie-rtems4.12, nios2-rtems4.12, or1k-rtems4.12, powerpc-rtems4.12, sh-rtems4.12, sparc-rtems4.12, sparc64-rtems4.12, v850-rtems4.12 2016-05-06T07:56:59 Board Support Package : arm-rtems4.12-xilinx_zynq_a9_qemu 2016-05-06T07:57:01 Checking for program 'arm-rtems4.12-gcc' : /build/rtems-4.12/bin/arm-rtems4.12-gcc 2016-05-06T07:57:02 Checking for program 'arm-rtems4.12-g++' : /build/rtems-4.12/bin/arm-rtems4.12-g++ 2016-05-06T07:57:04 Checking for program 'arm-rtems4.12-gcc' : /build/rtems-4.12/bin/arm-rtems4.12-gcc 2016-05-06T07:57:05 Checking for program 'arm-rtems4.12-ld' : /build/rtems-4.12/bin/arm-rtems4.12-ld 2016-05-06T07:57:07 Checking for program 'arm-rtems4.12-ar' : /build/rtems-4.12/bin/arm-rtems4.12-ar 2016-05-06T07:57:08 Checking for program 'arm-rtems4.12-nm' : /build/rtems-4.12/bin/arm-rtems4.12-nm 2016-05-06T07:57:10 Checking for program 'arm-rtems4.12-objdump' : /build/rtems-4.12/bin/arm-rtems4.12-objdump 2016-05-06T07:57:11 Checking for program 'arm-rtems4.12-objcopy' : /build/rtems-4.12/bin/arm-rtems4.12-objcopy 2016-05-06T07:57:13 Checking for program 'arm-rtems4.12-readelf' : /build/rtems-4.12/bin/arm-rtems4.12-readelf 2016-05-06T07:57:14 Checking for program 'arm-rtems4.12-strip' : /build/rtems-4.12/bin/arm-rtems4.12-strip 2016-05-06T07:57:16 Checking for program 'rtems-ld' : /build/rtems-4.12/bin/rtems-ld 2016-05-06T07:57:17 Checking for program 'rtems-tld' : /build/rtems-4.12/bin/rtems-tld 2016-05-06T07:57:19 Checking for program 'rtems-bin2c' : /build/rtems-4.12/bin/rtems-bin2c 2016-05-06T07:57:20 Checking for program 'tar' : /bin/tar 2016-05-06T07:57:22 Checking for program 'gcc, cc' : /build/rtems-4.12/bin/arm-rtems4.12-gcc 2016-05-06T07:57:23 Checking for program 'ar' : /build/rtems-4.12/bin/arm-rtems4.12-ar 2016-05-06T07:57:25 Checking for program 'g++, c++' : /build/rtems-4.12/bin/arm-rtems4.12-g++ 2016-05-06T07:57:26 Checking for program 'gas, gcc' : /build/rtems-4.12/bin/arm-rtems4.12-gcc 2016-05-06T07:57:28 Compiler version (arm-rtems4.12-gcc) : 6.0.1 20160415 (RTEMS 4.12, RSB 8bcc58b249fbf1f6e7bbd4151d9cf9b04c94b87e, Newlib 2.4.0) 2016-05-06T07:57:29 Checking for RTEMS CPU options header : ok 2016-05-06T07:57:31 Checking for RTEMS_DEBUG : no 2016-05-06T07:57:32 Checking for RTEMS_MULTIPROCESSING : no 2016-05-06T07:57:34 Checking for RTEMS_NEWLIB : yes 2016-05-06T07:57:35 Checking for RTEMS_POSIX_API : yes 2016-05-06T07:57:37 Checking for RTEMS_SMP : no 2016-05-06T07:57:38 Checking for RTEMS_NETWORKING : no 2016-05-06T07:57:40 Checking for RTEMS header : ok 2016-05-06T07:57:41 Checking for header dlfcn.h : yes 2016-05-06T07:57:43 Checking for header rtems/pci.h : yes 2016-05-06T07:57:44 'configure' finished successfully (1.176s) 2016-05-06T07:57:46 self:/scratch/git-rtems-libbsd (upstream) > waf 2016-05-06T07:57:47 Waf: Entering directory `/scratch/git-rtems-libbsd/build/arm-rtems4.12-xilinx_zynq_a9_qemu' 2016-05-06T07:57:49 [ 2/851] Compiling testsuite/include/rtems/bsd/test/network-config.h.in 2016-05-06T07:57:50 [ 2/851] Compiling rtemsbsd/rtems/generate_kvm_symbols 2016-05-06T07:57:52 [ 3/851] Compiling rtemsbsd/rtems/rtems-kvm-symbols.c 2016-05-06T07:57:53 [ 4/851] Compiling freebsd/contrib/libpcap/scanner.c 2016-05-06T07:57:55 [ 5/851] Compiling freebsd/lib/libc/net/nslexer.c 2016-05-06T07:57:56 [849/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/log01.exe 2016-05-06T07:57:58 [850/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/ftpd01.exe 2016-05-06T07:57:59 [851/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/ping01.exe 2016-05-06T07:58:01 Waf: Leaving directory `/scratch/git-rtems-libbsd/build/arm-rtems4.12-xilinx_zynq_a9_qemu' 2016-05-06T07:58:02 'build-arm-rtems4.12-xilinx_zynq_a9_qemu' finished successfully (52.713s) 2016-05-06T07:58:04 self:/scratch/git-rtems-libbsd (upstream) > waf 2016-05-06T07:58:05 Waf: Entering directory `/scratch/git-rtems-libbsd/build/arm-rtems4.12-xilinx_zynq_a9_qemu' 2016-05-06T07:58:07 [789/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/arphole.exe 2016-05-06T07:58:08 [791/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/commands01.exe 2016-05-06T07:58:10 [794/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/condvar01.exe 2016-05-06T07:58:11 [795/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/dhcpcd01.exe 2016-05-06T07:58:13 [797/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/dhcpcd02.exe 2016-05-06T07:58:14 [800/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/foobarclient.exe 2016-05-06T07:58:16 [801/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/foobarserver.exe 2016-05-06T07:58:17 [803/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/ftpd01.exe 2016-05-06T07:58:19 [805/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/init01.exe 2016-05-06T07:58:20 [807/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/lagg01.exe 2016-05-06T07:58:22 [809/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/log01.exe 2016-05-06T07:58:22 You will get kicked off the channel for the whole paste 2016-05-06T07:58:23 [811/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/loopback01.exe 2016-05-06T07:58:25 [834/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/media01.exe 2016-05-06T07:58:26 [835/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/mutex01.exe 2016-05-06T07:58:28 [836/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/netshell01.exe 2016-05-06T07:58:29 [837/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/ping01.exe 2016-05-06T07:58:31 [838/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/ppp01.exe 2016-05-06T07:58:32 [839/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/rwlock01.exe 2016-05-06T07:58:34 [840/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/selectpollkqueue01.exe 2016-05-06T07:58:35 [841/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/sleep01.exe 2016-05-06T07:58:37 [842/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/smp01.exe 2016-05-06T07:58:38 [843/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/swi01.exe 2016-05-06T07:58:40 [844/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/syscalls01.exe 2016-05-06T07:58:41 [845/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/telnetd01.exe 2016-05-06T07:58:43 [846/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/thread01.exe 2016-05-06T07:58:44 [847/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/timeout01.exe 2016-05-06T07:58:46 [848/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/unix01.exe 2016-05-06T07:58:46 sebhub: I need to head home now, sorry 2016-05-06T07:58:47 [849/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/usb01.exe 2016-05-06T07:58:49 [850/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/vlan01.exe 2016-05-06T07:58:50 *** peerst has quit IRC 2016-05-06T07:58:50 [851/851] Linking build/arm-rtems4.12-xilinx_zynq_a9_qemu/zerocopy01.exe 2016-05-06T07:58:52 Waf: Leaving directory `/scratch/git-rtems-libbsd/build/arm-rtems4.12-xilinx_zynq_a9_qemu' 2016-05-06T07:58:53 'build-arm-rtems4.12-xilinx_zynq_a9_qemu' finished successfully (4.221s) 2016-05-06T07:58:54 *** kiwichris_ has quit IRC 2016-05-06T07:58:57 *** peerst_ is now known as peerst 2016-05-06T08:00:01 sebhub: what's your question exactly? 2016-05-06T08:01:03 I build the libbsd using "waf", I change nothing, run "waf" again and it links the library and all executables again 2016-05-06T08:01:28 sebhub: ok, is any of the output files present in the source directory? 2016-05-06T08:05:14 the object files and libraries, no, they are in the build dir 2016-05-06T08:05:23 other files, maybe, I have to check 2016-05-06T08:06:20 sebhub: can you run "waf -v" and see if there is any warning displayed? 2016-05-06T08:07:19 i deleted the source tree, and checked it out from git again 2016-05-06T08:07:21 no warnings 2016-05-06T08:08:16 sebhub: and this is with waf 1.8.20 right? what's the URL to clone from git? 2016-05-06T08:08:23 a "waf -v" doesn't link the libbsd.a, a "waf" does 2016-05-06T08:08:37 waf --version 2016-05-06T08:08:39 waf 1.8.20 (14c787c991e1d9badd9d302bc9eaa060e9d4b03c) 2016-05-06T08:09:07 the "waf -v" links only the executables, a "waf" the libbsd.a and the executables 2016-05-06T08:13:00 sebhub: clone URL + commands you used please 2016-05-06T08:13:46 what do you mean with clone URL? 2016-05-06T08:14:02 sebhub: i do not have your repository and i do not know how to get it 2016-05-06T08:14:19 oh, thats a bit difficult to set up 2016-05-06T08:14:40 you need an rtems bsp and the cross compilers installled 2016-05-06T08:15:04 git://git.rtems.org/rtems-libbsd.git 2016-05-06T08:16:39 sebhub: is that the correct branch too? 2016-05-06T08:17:49 yes, the master 2016-05-06T08:18:39 sebhub: are the commands in README.waf up-to-date or are there other commands to use? 2016-05-06T08:19:56 the README.waf seems to be up to date 2016-05-06T08:20:07 waf configure --prefix=/build/rtems-4.12 --rtems-bsps=arm/xilinx_zynq_a9_qemu 2016-05-06T08:20:11 is what I used 2016-05-06T08:22:33 there is a typo in the README.waf file - "souce-builder" in step 1 2016-05-06T08:23:46 ok, it is not up to date, the 4.11 needs to be 4.12, I will fix this 2016-05-06T08:24:36 -> ../source-builder/sb-set-builder --prefix=/opt/rtems/4.12/tools 4.12/rtems-arm 2016-05-06T08:26:44 [git] 4835d26 by Sebastian Huber: [README.waf: Typo] http://devel.rtems.org/changeset/4835d26/rtems-libbsd 2016-05-06T08:26:54 [git] ff6534b by Sebastian Huber: [README.waf: Master branch must use 4.12] http://devel.rtems.org/changeset/ff6534b/rtems-libbsd 2016-05-06T08:30:21 ftp downloads are fine, but i do not see where checksums are verified 2016-05-06T08:30:49 ftp downloads of what? 2016-05-06T08:31:01 source-builder dependencies 2016-05-06T08:31:15 sorry, it takes a while to build those :-) 2016-05-06T08:31:18 the things the source builder downloads? 2016-05-06T08:31:25 yes 2016-05-06T08:32:00 they are verified with checksums, unfortunately this is optional 2016-05-06T08:32:35 https://devel.rtems.org/ticket/2536 2016-05-06T09:04:10 sebhub: on step 2, what's the current working directory? "in this example the path is /opt/rtems/4.12/bsps" 2016-05-06T09:04:42 sebhub: i have no "rtems-4.12" directory anywhere 2016-05-06T09:06:20 one moment 2016-05-06T09:10:40 [git] b39854a by Sebastian Huber: [README.waf: Add steps to clone and bootstrap RTEMS] http://devel.rtems.org/changeset/b39854a/rtems-libbsd 2016-05-06T09:11:09 please pull libbsd and use the updated README.waf 2016-05-06T09:11:22 thanks a lot for doing this 2016-05-06T09:21:16 sebhub: on step 5, do i need to create /opt/rtems/4.12/bsps and cd to it? 2016-05-06T09:22:12 sebhub: if i create b-realview_pbx_a9_qemu from rtems-4.12 the command ../rtems-4.12/configure is not found, but ../configure would work 2016-05-06T09:22:49 ok, this is a bit misleading 2016-05-06T09:23:03 all steps assume that you are in one common directory 2016-05-06T09:24:38 ideally, there would be a single annotated build.sh to run all this ? 2016-05-06T09:25:11 ok, good idea 2016-05-06T09:26:11 it will take some time for me to do this, e.g. building the tool chain, etc. 2016-05-06T09:26:32 it is just about turning what's in the README into a list of commands :-) 2016-05-06T09:26:42 else i can keep asking questions 2016-05-06T09:30:13 I have to figure out how these annotations work with asciidoc 2016-05-06T09:31:35 after step 1, i had a PATH environment variable pointing to something so i decided to start over 2016-05-06T09:32:09 on step 4, i am getting lots of "autoreconf: command not found" 2016-05-06T09:33:45 i also had to make /opt/rtems/4.12/tools writable for my user by the way, so it may be worth mentioning something about this 2016-05-06T09:35:26 (starting over again...) 2016-05-06T09:36:17 I have to discuss this with chris 2016-05-06T10:20:22 sebhub: bootstrapping the RTEMS sources (step 4) requires the binaries that were built in source-builder, does it not? Yet there is no PATH modification between steps 2 and 4 2016-05-06T10:25:01 sebhub: PATH=/opt/rtems/4.12/tools/bin/:$PATH ./bootstrap seems to work 2016-05-06T10:28:04 yes, it needs the autotools built by RSB 2016-05-06T10:28:28 we need a ready to run script, I guess 2016-05-06T10:29:03 I have to discuss with Chris how we set up the various directories 2016-05-06T10:29:17 you can install everything in your $HOME for example 2016-05-06T10:37:13 ok, i had to redo step 5 because i did not pass PATH on the command-line the first time, and then it failed to pick-it up so i removed the whole folder and started over 2016-05-06T10:37:43 sebhub: i have managed to build the project; calling waf again does not cause any re-link of the exe files 2016-05-06T10:45:56 sebhub: the build i have is arm-rtems4.12-realview_pbx_a9_qemu which is not arm-rtems4.12-xilinx_zynq_a9_qemu 2016-05-06T10:50:30 hm, I will start from scratch and check if this solves the problem 2016-05-06T10:50:56 i am building xilinx_zynq_a9_qemu now 2016-05-06T10:51:17 this should have no influence 2016-05-06T10:51:46 yes, no change, and no rebuild when calling waf a second time 2016-05-06T10:53:26 sebhub: https://pastebin.mozilla.org/8869828 <- 2016-05-06T10:55:16 sebhub: can you still reproduce the problem? can you tar up your folder and expose it somewhere? 2016-05-06T10:57:04 hm, it has 7.3GiB 2016-05-06T10:57:16 the libbsd folder?? 2016-05-06T10:57:31 yes, i have severl git submodules 2016-05-06T10:57:36 i see 2016-05-06T10:58:09 after waf clean it my folder takes ~500MB :-( 2016-05-06T10:58:30 ls 2016-05-06T10:58:53 oh, clean does not clean the previous bsps 2016-05-06T11:00:16 I modified the steps to be a single shell script 2016-05-06T11:00:26 I try to execute the steps now 2016-05-06T11:01:11 Chris doesn't have these link problems, I guess this has something to do with my local directory 2016-05-06T11:02:20 sebhub: yes, though such things never happen unless build files are present in the source directory so i would like to understand how you managed to do such at thing 2016-05-06T11:03:00 my directory is several years old and it pre-dates the waf support 2016-05-06T11:06:20 I did a "rm -rf rtemsbsd freebsd ; git co ." though, these should be the source dirs 2016-05-06T11:07:42 sebhub: what happens when running "waf clean build build" ? 2016-05-06T11:12:08 https://pastebin.mozilla.org/8869832 2016-05-06T11:12:51 the "build build" seems to not work properly 2016-05-06T11:12:55 this is an unrelated bug in the build scripts :-) 2016-05-06T11:13:50 "waf" and "waf build" behave differently 2016-05-06T11:14:26 https://pastebin.mozilla.org/8869833 2016-05-06T11:14:34 a "waf" links also libbsd.a 2016-05-06T11:17:51 the exception is unrelated and "waf build" does the same thing as "waf" 2016-05-06T11:18:12 there is a hidden state somewhere that prevents running consecutive builds in the same process 2016-05-06T11:21:25 in rtems_waf/rtems.py there is an extra line "@TaskGen.extension('.c')" which does not make sense 2016-05-06T11:22:56 I have no idea how this waf stuff works, this is all the work from Chris 2016-05-06T11:23:46 sebhub: that piece of code looks wrong, does it not? 2016-05-06T11:25:21 and the reason for the exception is that get_env comes from rtems_waf.rtems.context the first time and from waflib.Build.BuildContext the second 2016-05-06T11:25:21 *** printk has joined #rtems 2016-05-06T11:28:52 sebhub: https://pastebin.mozilla.org/8869834 2016-05-06T11:29:41 *** ita has quit IRC 2016-05-06T11:29:41 *** ita has joined #rtems 2016-05-06T11:45:02 [git] cf72971 by Sebastian Huber: [README.waf: Make set-up steps self contained] http://devel.rtems.org/changeset/cf72971/rtems-libbsd 2016-05-06T11:45:24 I hopefully fixed the README.waf 2016-05-06T11:46:15 if I execute the set-up steps in a fresh directory, then the re-link is gone, so this must have something to do with my local directory 2016-05-06T12:11:22 [git] 469db42 by Sebastian Huber: [subr_prf.c: Import from FreeBSD] http://devel.rtems.org/changeset/469db42/rtems-libbsd 2016-05-06T12:11:22 [git] d4402dd by Sebastian Huber: [bcd.c: Import from FreeBSD] http://devel.rtems.org/changeset/d4402dd/rtems-libbsd 2016-05-06T12:11:22 [git] 54a577d by Sebastian Huber: [subr_prf.c: Port to RTEMS] http://devel.rtems.org/changeset/54a577d/rtems-libbsd 2016-05-06T12:11:22 [git] 356f59c by Sebastian Huber: [kvprintf: Add support for %m] http://devel.rtems.org/changeset/356f59c/rtems-libbsd 2016-05-06T12:11:22 [git] b8f127e by Sebastian Huber: [Update kernel namespace] http://devel.rtems.org/changeset/b8f127e/rtems-libbsd 2016-05-06T12:40:43 [git] f9a697a by Sebastian Huber: [media01: Add sysctl command] http://devel.rtems.org/changeset/f9a697a/rtems-libbsd 2016-05-06T12:40:43 [git] edb5e4b by Sebastian Huber: [Add kernel namespace exceptions These names correspond to sysctls.] http://devel.rtems.org/changeset/edb5e4b/rtems-libbsd 2016-05-06T12:45:18 *** kiwichris has joined #rtems 2016-05-06T12:46:54 *** printk has quit IRC 2016-05-06T12:47:08 kiwichris: hi there 2016-05-06T12:47:20 Hi 2016-05-06T12:47:26 kiwichris: https://pastebin.mozilla.org/8869834 2016-05-06T12:49:46 [git] 81e5ff4 by Sebastian Huber: [Add kernel namespace exception for mbstat] http://devel.rtems.org/changeset/81e5ff4/rtems-libbsd 2016-05-06T12:52:27 *** printk has joined #rtems 2016-05-06T12:52:28 ita, what is meant by specified multiple times? 2016-05-06T12:52:50 kiwichris: try running "waf build build clean build" 2016-05-06T12:55:01 ita, I get an error in libbsd_waf.py 2016-05-06T12:55:37 kiwichris: the error is because the replacement is incomplete 2016-05-06T12:55:56 kiwichris: so the first build gets an rtems context, while the second gets a normal buildcontext 2016-05-06T12:56:28 Oh and the default env returned is a list 2016-05-06T12:56:39 kiwichris: an empty list even 2016-05-06T12:56:49 Yes 2016-05-06T12:57:11 Do I need to parse the Options.commands list better and handle each case? 2016-05-06T12:58:46 kiwichris: i think so, or at least make a note somewhere in case someone else asks in the future? 2016-05-06T12:59:18 kiwichris: And for the @extension line, i think it makes no sense there. Is that a copy-paste leftover? 2016-05-06T12:59:19 I will fix the code. Thanks for the review and feedback. 2016-05-06T12:59:58 It could be 2016-05-06T13:00:22 i was trying to understand what sebhub did to get the errors he found; a few changes were needed in README.waf 2016-05-06T13:00:24 I will remove it. 2016-05-06T13:01:05 I did not see the error. I had to head out to meet people and I am on the road early next week so had to kill the connection. 2016-05-06T13:05:24 *** kiwichris has quit IRC 2016-05-06T13:13:01 *** kiwichris has joined #rtems 2016-05-06T13:18:53 *** sebhub has quit IRC 2016-05-06T13:33:51 *** monstr has quit IRC 2016-05-06T13:34:44 *** printk has quit IRC 2016-05-06T13:40:44 *** dougk_ has joined #rtems 2016-05-06T14:07:44 *** jkridner|work has joined #rtems 2016-05-06T14:08:32 *** jkridner has quit IRC 2016-05-06T15:55:39 *** jkridner|work has quit IRC 2016-05-06T15:56:09 *** jkridner has joined #rtems 2016-05-06T16:01:03 *** jkridner has quit IRC 2016-05-06T17:06:37 *** dougk_ has quit IRC 2016-05-06T17:14:23 *** shelf has quit IRC 2016-05-06T17:16:28 *** shelf has joined #rtems 2016-05-06T17:48:01 *** ambro718 has joined #rtems 2016-05-06T17:51:26 *** shelf has quit IRC 2016-05-06T17:55:03 *** shelf has joined #rtems 2016-05-06T17:58:50 *** darnir has quit IRC 2016-05-06T18:00:27 *** darnir has joined #rtems 2016-05-06T18:03:09 *** shelf has quit IRC 2016-05-06T18:10:37 *** shelf has joined #rtems 2016-05-06T18:17:26 *** shelf has quit IRC 2016-05-06T18:19:41 *** shelf has joined #rtems 2016-05-06T18:40:57 *** kiwichris has quit IRC 2016-05-06T18:41:09 *** kiwichris has joined #rtems 2016-05-06T19:24:55 *** mkhoory has quit IRC 2016-05-06T19:46:52 [git] 5b1f20b by Sebastian Huber: [Rename files for kernel namespace script This makes it easier to create the kernel namespace header.] http://devel.rtems.org/changeset/5b1f20b/rtems-libbsd 2016-05-06T19:46:52 [git] 5b4394f by Sebastian Huber: [Add namespace support for kvm] http://devel.rtems.org/changeset/5b4394f/rtems-libbsd 2016-05-06T20:12:35 [git] b4d7d5d by Martin Galvan: [sonic.c: Fix warning on gcc 6 GCC 6.0 previews give the following warning on sonic.c: ../../../../../rtems/c/src/libchip/network/sonic.c:837:11: warning: statement is indented as if it were guarded by... [-Wmisleading-indentation] This was discussed here: https://lists.rtems.org/pipermail/devel/2016-March/014004.html and we 2016-05-06T20:12:35 concluded it would be bet...] http://devel.rtems.org/changeset/b4d7d5d/rtems 2016-05-06T20:17:45 *** gedare has joined #rtems 2016-05-06T20:17:45 *** ChanServ sets mode: +o gedare 2016-05-06T20:46:06 *** ambro718 has quit IRC 2016-05-06T20:49:27 *** gedare has quit IRC 2016-05-06T21:10:51 *** mumptai has joined #rtems 2016-05-06T21:57:54 *** kiwichris has quit IRC 2016-05-06T21:58:26 *** kiwichris has joined #rtems 2016-05-07T02:23:42 *** lcpfnvcy has quit IRC 2016-05-07T02:47:30 *** mumptai has quit IRC 2016-05-07T02:59:42 *** mumptai has joined #rtems 2016-05-07T03:16:34 *** kiwichris has quit IRC 2016-05-07T03:17:21 *** kiwichris has joined #rtems 2016-05-07T05:30:09 *** kiwichris has quit IRC 2016-05-07T05:32:18 *** kiwichris has joined #rtems 2016-05-07T05:35:08 [git] 44a82d0 by Chris Johns: [Update rtems_waf with the latest fixes.] http://devel.rtems.org/changeset/44a82d0/rtems-libbsd 2016-05-07T05:47:06 *** kiwichris has quit IRC 2016-05-07T05:47:41 *** kiwichris has joined #rtems 2016-05-07T06:50:26 *** kiwichri_ has joined #rtems 2016-05-07T06:50:34 *** kiwichris has quit IRC 2016-05-07T07:01:33 *** lcpfnvcy has joined #rtems 2016-05-07T07:01:43 *** ambro718 has joined #rtems 2016-05-07T07:44:45 *** kiwichri_ has quit IRC 2016-05-07T07:44:45 *** kiwichris has joined #rtems 2016-05-07T08:58:01 *** kiwichris has quit IRC 2016-05-07T08:58:39 *** kiwichris has joined #rtems 2016-05-07T09:15:45 *** kiwichri_ has joined #rtems 2016-05-07T09:16:15 *** kiwichris has quit IRC 2016-05-07T10:19:30 *** kiwichri_ has quit IRC 2016-05-07T10:20:08 *** kiwichris has joined #rtems 2016-05-07T10:37:58 *** kiwichris has quit IRC 2016-05-07T10:38:15 *** kiwichris has joined #rtems 2016-05-07T15:36:03 *** kiwichri_ has joined #rtems 2016-05-07T15:36:32 *** kiwichris has quit IRC 2016-05-07T16:46:19 *** verm__ has quit IRC 2016-05-07T16:46:28 *** verm__ has joined #rtems 2016-05-07T18:22:21 *** mumptai has quit IRC 2016-05-07T18:23:23 *** mumptai has joined #rtems 2016-05-07T19:16:42 *** kiwichris has joined #rtems 2016-05-07T19:16:47 *** kiwichri_ has quit IRC 2016-05-07T22:04:37 *** bbhoss has quit IRC 2016-05-07T22:05:01 *** bbhoss has joined #rtems 2016-05-07T22:22:50 *** kiwichri_ has joined #rtems 2016-05-07T22:23:06 *** kiwichris has quit IRC 2016-05-07T22:26:45 *** mumptai has quit IRC 2016-05-07T22:37:09 *** ambro718 has quit IRC 2016-05-07T23:30:56 *** kiwichri_ has quit IRC 2016-05-07T23:31:01 *** kiwichris has joined #rtems 2016-05-07T23:36:43 *** kiwichris has quit IRC 2016-05-08T00:27:40 *** ita has quit IRC 2016-05-08T01:15:31 *** ita has joined #rtems 2016-05-08T06:58:21 *** ambro718 has joined #rtems 2016-05-08T08:55:41 *** kiwichris has joined #rtems 2016-05-08T09:06:50 *** kiwichris has quit IRC 2016-05-08T11:49:47 *** shelf is now known as timothyh 2016-05-08T12:26:22 *** ambro718 has quit IRC 2016-05-08T17:22:50 *** ita has quit IRC 2016-05-08T18:47:04 *** ita has joined #rtems 2016-05-08T19:11:57 *** mumptai has joined #rtems 2016-05-08T22:20:11 *** mumptai has quit IRC 2016-05-08T22:45:23 *** kiwichris has joined #rtems 2016-05-08T22:54:04 *** kiwichris has quit IRC