Final GSoC submission

Hi, Apparently we're not allowed to be evaluated on anything we submit after today at noon PDT. I thought with that in mind I'd send in my submission. Final status update * Wrote docs * Cleaned up a bit, removed dead code * mach_wait_probe now handles the size of the memory returned,

Re: Wiki repository now available though Savannah

Hi, On Mon, Aug 4, 2008 at 10:31 PM, <[EMAIL PROTECTED]> wrote: > Hi, > > On Mon, Aug 04, 2008 at 10:46:55AM +0200, Thomas Schwinge wrote: >> On Mon, Aug 04, 2008 at 03:00:46AM +0200, [EMAIL PROTECTED] >> wrote: > >> This is done now: >> > > B

Re: GSoC projects

Hi, Sorry for the delay in answering. Due to Comcast's incompetence I don't have internet access at home until next week. a) Instrumentation and a SystemTap port. b) The instrumentation patches will make their way into mach, the SystemTap work should probably go into its own git repository. T

[task #6537] Implement ddb/db_elf.c

Follow-up Comment #5, task #6537 (project hurd): Fixed the db_error issue and the line number bug. Same changelog. (file #16184) ___ Additional Item Attachment: File name: elf-patch Size:111 KB _

[task #6537] Implement ddb/db_elf.c

Follow-up Comment #4, task #6537 (project hurd): 2008-07-18 Andrei Barbu <[EMAIL PROTECTED]> * ddb/db_elf.c: New file * ddb/db_sym.c: ELF prototypes and switch entries * ddb/db_sym.h: ELF symbol table type as the default * ddb/dwarf2.h: New file

[patch #6583] db_read_bytes reports an incorrect address in its error message

on box: ___ Details: 2008-07-18 Andrei Barbu <[EMAIL PROTECTED]> * i386/i386/db_interface.c(db_read_bytes): Fix the error message address ___ File

[patch #6582] ddb doesn't decode arguments for aam and aad instructions

cussion box: ___ Details: 2008-07-18 Andrei Barbu <[EMAIL PROTECTED]> * i386/i386/db_disasm.c (db_disasm): Decode operands for aam and aad instructions ___

[task #6537] Implement ddb/db_elf.c

Follow-up Comment #3, task #6537 (project hurd): This is a rewrite of the elf backend to use dwarf instead of the elf symbols. It adds many features, the two most obvious ones are that it's a lot slower and that line numbers and files work in ddb. This is also a requirement for probe support. (f

[patch #6581] Add posix types

etails: 2008-07-18 Andrei Barbu <[EMAIL PROTECTED]> * include/sys/types.h: Add posix types ___ File Attachments: --- Date: Friday 07/18/2008 at 17:22 Name: posix-types

[patch #6580] ddb still prints empty symbols

on box: ___ Details: Instead of an empty symbol print the address, there's no point in that empty string. 2008-07-18 Andrei Barbu <[EMAIL PROTECTED]> * ddb/db_sym.c (db_task_printsym): Don't print empty symbols ___

[patch #6579] Remove quad from sys/types.h

on box: ___ Details: 2008-07-18 Andrei Barbu <[EMAIL PROTECTED]> * include/sys/types.h: Get rid of quad ___ File Attachments: --- Date: Friday 07/18/2008 at 17:18 Name:

[patch #6578] db_set_breakpoint should return the breakpoint, and add a new breakpoint flag

on box: ___ Details: For kernel probes db_set_breakpoint has to return the breakpoint it allocated. A new breakpoint type is needed to let DDB know that it is external and not a regular breakpoint. 2008-07-18 Andrei Barbu <[EMAIL

[patch #6577] Free ddb symbols

on box: ___ Details: Symbols need to be freed for ELF; this adds the free_symbol entry to db_sym_switch, as well as frees symbols where they're used. This is a nop for anything but ELF, so at the moment it does nothing. 2008-07-18 Andrei Barbu <[EMAIL PROTECTED]> * ddb/db

Re: Revision control

If we're throwing ideas out. I suggest darcs. http://better-scm.berlios.de/comparison/comparison.html Looks at a whole host of features. Andrei

Re: dtrace/systemtap options

> I don't get this: Why implement module loading just for this? Can't > SystemTap be just statically linked in gnumach? Or are the actual test > functions compiled and loaded as modules in SystemTap?... > SystemTap itself doesn't get to sit in the Linux kernel. Linux exposes an in-kernel API calle

Re: Revision control

On Wed, May 28, 2008 at 5:37 PM, Michael Banck <[EMAIL PROTECTED]> wrote: > On Wed, May 28, 2008 at 10:34:41PM +0200, Arne Babenhauserheide wrote: >> I like the idea. >> >> It would lower the barrier to begin coding on the hurd (I think), and it >> would >> be a good opportunity to clean up the cv

Revision control

Hi, Would there be support for moving to a different revision control mechanism? Anything is far better than cvs, especially if it's distributed. I saw there was some talk of this a while ago but nothing happened. We've got lots of good options to pick from that we know will scale well. Andrei

[task #6537] Implement ddb/db_elf.c

Additional Item Attachment, task #6537 (project hurd): File name: changelog-elf-1.patch Size:1 KB ___ Reply to this item at: ___ Message sent

[task #6537] Implement ddb/db_elf.c

Follow-up Comment #2, task #6537 (project hurd): Updated the reflect the discussion on bug-hurd. (file #15746) ___ Additional Item Attachment: File name: elf-1.patchSize:71 KB

Re: dtrace/systemtap options

On Mon, May 26, 2008 at 5:27 PM, Samuel Thibault <[EMAIL PROTECTED]> wrote: > [EMAIL PROTECTED], le Mon 26 May 2008 21:25:33 +0200, a écrit : >> Maybe you actually mean that RW increases the risk of crashing?... >> (Well, what fun would there be in system development, without the >> prospect of reg

Re: [task #6537] Implement ddb/db_elf.c

> Ok, then put them into sys/types.h, since that's where POSIX says they > are :) >> > - I would have put __CONCAT in the ELF header, as it's the only place >> > where it is used. Other helper macros aren't used. >> Same as above. > > Then another place too, it really doesn't belong to vm_types.h

Re: [task #6537] Implement ddb/db_elf.c

Hey, Thanks for the comments. > - take care of indentation. Try to respect the same existing indentation > (yes it's not homogeneous, follow it per-file). For instance, in > new-gnumach-3/i386/i386at/model_dep.c Ah, sorry. My emacs automatically reindents every file to my settings, I'll disable

dtrace/systemtap options

Hi, I've been thinking about all of the various ways to get either Dtrace or SystemTap working. I've come up with 4 options: 1 Dtrace would need to sit entirely out of the kernel. About 3k lines of simple code to rewrite, FSF lawyers are ok with this option. Since it must sit in userland and it

[task #7179] Multiprocessor support

Follow-up Comment #1, task #7179 (project hurd): I don't think we can steal it from xnu. Xnu is under the Apple Public Source License, it's not GPL-compatible. ___ Reply to this item at: ___

[patch #6508] Building with -g3

URL: Summary: Building with -g3 Project: The GNU Hurd Submitted by: abarbu Submitted on: Friday 05/09/2008 at 15:45 Category: GNU Mach Priority: 5 - Normal

[patch #6507] Fixing delay

URL: Summary: Fixing delay Project: The GNU Hurd Submitted by: abarbu Submitted on: Friday 05/09/2008 at 15:40 Category: GNU Mach Priority: 5 - Normal

[task #6537] Implement ddb/db_elf.c

Follow-up Comment #1, task #6537 (project hurd): Two patches that add this. The first extends symbol_values to take the symbol table as an argument, this is required for ELF. The second adds db_elf.c as well as a whole bunch of support files, enables ELF and adds it to model_dep. It also sets ELF