Kdgdb Changes #12

ariscop · 2016-04-20T03:18:43Z

No description provided.

gigaherz · 2016-04-20T16:20:22Z

There's no need to close PRs, you can just force-push to the same branch, github is ok with that.

learn-more · 2016-04-20T18:41:39Z

reactos/lib/rtl/actctx.c

-            DPRINT1("Unsupported yet language attribute (%.*S)\n",
-                    attr_value.len, attr_value.ptr);
+            DPRINT("Unsupported yet language attribute (%.*S)\n",
+                   attr_value.len, attr_value.ptr);


This is a diff from upstream?

Yeah, i'm using the repo at git.reactos.org rather than github since it lags behind

ariscop · 2016-04-23T02:54:21Z

@gigaherz Wanted to change the branch name, minor but it bothered me

ariscop · 2016-04-23T06:50:32Z

I think the issue mentioned in 37445e0 is that it loads bss sections at the preferred base rather than relative to the text section, that will happen with pretty much everything in kernel space though, not just ntoskrnl

I'm hoping to pass section addresses out instead of base addresses to solve this

ariscop · 2016-04-28T03:55:47Z

I'm thinking there should be a mode where cores are represented as threads, Neither gdb or kdapi is prepared to handle debugging individual processes within a kernel

ariscop · 2016-04-29T13:44:12Z

I'm starting to think it isn't worth using kdapi, unless one really wants a gdb stub for windows
Adding some way to set KiDebugRoutine from kdgdb may prove simpler

Making function calls by Returning is terribly awkward

- Remove a debug message that should have been removed before committing. - Also fix a test failure. git-svn-id: file:///srv/svn/reactos/trunk@71244 97493ccd-5924-5043-b1f5-66cb403b36ce

The onliner: - Move to next entry by applying the offset to the previous entry, and not to the first entry This fixes enumerating ADS from files with 3+ streams. Previously, doing dir /R in a directory would have lead to a cmd crash. CORE-11164 #resolve #comment Fixed by r71245 git-svn-id: file:///srv/svn/reactos/trunk@71245 97493ccd-5924-5043-b1f5-66cb403b36ce

…t tests), by Sebastian Lackner & Me CORE-10923 git-svn-id: file:///srv/svn/reactos/trunk@71248 97493ccd-5924-5043-b1f5-66cb403b36ce

- Update to 0.46d7. CORE-11157 git-svn-id: file:///srv/svn/reactos/trunk@71252 97493ccd-5924-5043-b1f5-66cb403b36ce

git-svn-id: file:///srv/svn/reactos/trunk@71255 97493ccd-5924-5043-b1f5-66cb403b36ce

CORE-11161 git-svn-id: file:///srv/svn/reactos/trunk@71259 97493ccd-5924-5043-b1f5-66cb403b36ce

Mark the CORE-11057 fix as local change. CORE-11172 git-svn-id: file:///srv/svn/reactos/trunk@71260 97493ccd-5924-5043-b1f5-66cb403b36ce

…n opening an empty file. CORE-11166 git-svn-id: file:///srv/svn/reactos/trunk@71261 97493ccd-5924-5043-b1f5-66cb403b36ce

git-svn-id: file:///srv/svn/reactos/trunk@71262 97493ccd-5924-5043-b1f5-66cb403b36ce

…s when validating file times. Also silence some debug outputs in the normal usage flow. git-svn-id: file:///srv/svn/reactos/trunk@71263 97493ccd-5924-5043-b1f5-66cb403b36ce

Implement FsRtlValidateReparsePointBuffer(). CORE-11172 git-svn-id: file:///srv/svn/reactos/trunk@71264 97493ccd-5924-5043-b1f5-66cb403b36ce

Upgrade the WinBtrfs to release 0.4. CORE-11172 #resolve #Committed in r71265 git-svn-id: file:///srv/svn/reactos/trunk@71265 97493ccd-5924-5043-b1f5-66cb403b36ce

Romanian resources maintenance. Patch by Ștefan Fulea. CORE-11111 #resolve #comment Thanks a lot! git-svn-id: file:///srv/svn/reactos/trunk@71266 97493ccd-5924-5043-b1f5-66cb403b36ce

Updating Turkish Translation. Patch by Erdem Ersoy. CORE-11130 #resolve #comment Thanks a lot! git-svn-id: file:///srv/svn/reactos/trunk@71267 97493ccd-5924-5043-b1f5-66cb403b36ce

Fix serious path localization issue. Patch by Wim Hueskes. CORE-9656 #resolve #comment Thanks a lot! git-svn-id: file:///srv/svn/reactos/trunk@71268 97493ccd-5924-5043-b1f5-66cb403b36ce

Add the romanian resource to the repository. CORE-11111 #resolve Fix build. git-svn-id: file:///srv/svn/reactos/trunk@71269 97493ccd-5924-5043-b1f5-66cb403b36ce

In case we're performing an in-place move (ie, in the same directory), if short name matches before and after, keep it. This avoids that a file with long file name sees its short name changing on normally 'non-effect' renames. For instance, in case of case-change only rename, short name will now be preserved. See ya, you 7y old bug :-) CORE-3875 #resolve #comment Fixed with r71270 git-svn-id: file:///srv/svn/reactos/trunk@71270 97493ccd-5924-5043-b1f5-66cb403b36ce

… Calvo. CID 1341307 CORE-11177 #resolve #comment Thanks! git-svn-id: file:///srv/svn/reactos/trunk@71271 97493ccd-5924-5043-b1f5-66cb403b36ce

…tínez Calvo. CID 1322300 CORE-11181 #resolve #comment Thanks! git-svn-id: file:///srv/svn/reactos/trunk@71272 97493ccd-5924-5043-b1f5-66cb403b36ce

Ease adding the remaining vCont types

these are the same as other queries

missed this somehow >_<

Can now disconnect/reconnect gdb without issue

Windows has this funny property where the lower two bits of handles are ignored, so we can safely use pid/tid 1 to represent the idle process, rather than shifting every pid

To make continue work agaun run: set schedule-multiple on set scheduler-locking off

Can now load all kernel symbols just by running load-symbols This is less useful for kdgdb and more for when you're attached to qemu, or some hardware debug interface

This reduces the amount of boilerplate, and ensures that the current thread is always valid Moving init until phase 1 because before then a large number of kernel structures are unitialized, such as the process list. By phase 1 everythings sane

Use with info w32 tib Note that this crashes gdb 7.11, bug reported upstream

Not yet quite working as gdb issues a single-step after a breakpoint, which isn't implemented yet

Seems more accurate

Correct "E" replies to "E\x\x", and handle ack/nak responses correctly

this code was triggering a bugcheck and it was easier to rewrite than fix, plus it removes the buffer (gdb accepts infinite length packets) and generally is nicer The special handling for the idle thread is gone too, it's not clear why it was done but pid/tid 0 seems to just be an alias for 4.8

This changes whether processes are represented as individual inferiors, tuning this off treats the kernel as a single inferior with many threads. Useful to avoid bugs in current gdb Command line option to be added later

Currently it appears impossible to relocate the current executable in gdb https://sourceware.org/bugzilla/show_bug.cgi?id=20007 Also disable multiprocess mode by default, since it breaks upstream gdb

ariscop · 2016-05-09T03:01:37Z

I have single step 'working' but it so readily crashes both reactos and gdb

ariscop · 2016-06-06T09:55:37Z

Stopped working on this, gdb just can't handle it and kdapi really isn't built for this kinda use

Adding kd support to gdb would be a more productive and workable goal

AmineKhaldi · 2016-09-14T10:13:43Z

Can this be adapted to the recent work that has been done on kdgdb?

ariscop mentioned this pull request Apr 20, 2016

Always build kdgdb #11

Closed

learn-more reviewed Apr 20, 2016
View reviewed changes

ariscop force-pushed the kdgdb branch 4 times, most recently from dfeb17b to d55936b Compare April 26, 2016 13:51

gadamopoulos and others added 19 commits May 3, 2016 17:10

[SHELL32]

fdb3129

- Remove a debug message that should have been removed before committing. - Also fix a test failure. git-svn-id: file:///srv/svn/reactos/trunk@71244 97493ccd-5924-5043-b1f5-66cb403b36ce

[WINTRUST_WINETEST] Import Wine Staging commit bdb15cf (WinVerifyTrus…

8e247bd

…t tests), by Sebastian Lackner & Me CORE-10923 git-svn-id: file:///srv/svn/reactos/trunk@71248 97493ccd-5924-5043-b1f5-66cb403b36ce

[UNIATA]

bba89df

- Update to 0.46d7. CORE-11157 git-svn-id: file:///srv/svn/reactos/trunk@71252 97493ccd-5924-5043-b1f5-66cb403b36ce

[UNIATA] Fix PCH.

c38d94e

git-svn-id: file:///srv/svn/reactos/trunk@71255 97493ccd-5924-5043-b1f5-66cb403b36ce

[FREELDR] fix copypasta

14a64d4

CORE-11161 git-svn-id: file:///srv/svn/reactos/trunk@71259 97493ccd-5924-5043-b1f5-66cb403b36ce

[BTRFS]

7693a85

Mark the CORE-11057 fix as local change. CORE-11172 git-svn-id: file:///srv/svn/reactos/trunk@71260 97493ccd-5924-5043-b1f5-66cb403b36ce

[NTOSKRNL][NTDLL_APITEST] Fix error returned from NtCreateSection whe…

55903ab

…n opening an empty file. CORE-11166 git-svn-id: file:///srv/svn/reactos/trunk@71261 97493ccd-5924-5043-b1f5-66cb403b36ce

fix the build for msvc (2013?) users.

62f51fe

git-svn-id: file:///srv/svn/reactos/trunk@71262 97493ccd-5924-5043-b1f5-66cb403b36ce

[NTOSKRNL] Fix NtApphelpCacheControl image lookup failure return code…

5f573e4

…s when validating file times. Also silence some debug outputs in the normal usage flow. git-svn-id: file:///srv/svn/reactos/trunk@71263 97493ccd-5924-5043-b1f5-66cb403b36ce

[NTOSKRNL_VISTA]

c5908d5

Implement FsRtlValidateReparsePointBuffer(). CORE-11172 git-svn-id: file:///srv/svn/reactos/trunk@71264 97493ccd-5924-5043-b1f5-66cb403b36ce

[BTRFS]

0b4ca15

Upgrade the WinBtrfs to release 0.4. CORE-11172 #resolve #Committed in r71265 git-svn-id: file:///srv/svn/reactos/trunk@71265 97493ccd-5924-5043-b1f5-66cb403b36ce

[TRANSLATION]

f1cbac2

Romanian resources maintenance. Patch by Ștefan Fulea. CORE-11111 #resolve #comment Thanks a lot! git-svn-id: file:///srv/svn/reactos/trunk@71266 97493ccd-5924-5043-b1f5-66cb403b36ce

[TRANSLATION]

822715b

Updating Turkish Translation. Patch by Erdem Ersoy. CORE-11130 #resolve #comment Thanks a lot! git-svn-id: file:///srv/svn/reactos/trunk@71267 97493ccd-5924-5043-b1f5-66cb403b36ce

[USETUP]

973136a

Fix serious path localization issue. Patch by Wim Hueskes. CORE-9656 #resolve #comment Thanks a lot! git-svn-id: file:///srv/svn/reactos/trunk@71268 97493ccd-5924-5043-b1f5-66cb403b36ce

[IERNONCE]

9e57239

Add the romanian resource to the repository. CORE-11111 #resolve Fix build. git-svn-id: file:///srv/svn/reactos/trunk@71269 97493ccd-5924-5043-b1f5-66cb403b36ce

[KBDTOOL] Handle failure to parse scancodes. Patch by Víctor Martínez…

58f7c27

… Calvo. CID 1341307 CORE-11177 #resolve #comment Thanks! git-svn-id: file:///srv/svn/reactos/trunk@71271 97493ccd-5924-5043-b1f5-66cb403b36ce

[UDFS] Avoid double free in UDFLoadLogicalVolInt. Patch by Víctor Mar…

510580b

…tínez Calvo. CID 1322300 CORE-11181 #resolve #comment Thanks! git-svn-id: file:///srv/svn/reactos/trunk@71272 97493ccd-5924-5043-b1f5-66cb403b36ce

ariscop added 25 commits May 8, 2016 18:32

Whoops

d89e053

Rearange handle_gdb_v

0b62cb6

Ease adding the remaining vCont types

Stub qThreadExtraInfo

6641428

Clean up thread-id formatting and switch thread handler

5f8788c

Handle 'Q' packets

fc7bc49

these are the same as other queries

Install kdgdb

6aac0aa

missed this somehow >_<

Enable extended-remote

cac55a5

Can now disconnect/reconnect gdb without issue

Map 0 to 1 instead of shifting all pid's

6c25c87

Windows has this funny property where the lower two bits of handles are ignored, so we can safely use pid/tid 1 to represent the idle process, rather than shifting every pid

Error unless gdb is asking for all threads to be resumed

f75e860

To make continue work agaun run: set schedule-multiple on set scheduler-locking off

Resend on reception error

ead00b5

Some .gdbinit nonsense

9a44d7f

Can now load all kernel symbols just by running load-symbols This is less useful for kdgdb and more for when you're attached to qemu, or some hardware debug interface

Implement qGetTIBAddr

cfd9608

Use with info w32 tib Note that this crashes gdb 7.11, bug reported upstream

Make send_gdb_packet take printf-style format

0e97f49

Implement breakpoints

0b07c80

Not yet quite working as gdb issues a single-step after a breakpoint, which isn't implemented yet

Use MmSystemRangeStart instead of MmHighestUserAddress

ce73f97

Seems more accurate

Slight load-symbols improvement

e40652f

PSTRING isn't an alias for char*

70d3528

Minor gdbinit cleanup

517cc6d

Improve link-layer

c5de062

Correct "E" replies to "E\x\x", and handle ack/nak responses correctly

Ternary operator~

9c512c3

Remove ptid_from_process

8dbc0d0

Add boolean to toggle multiprocess support

d9a7f15

This changes whether processes are represented as individual inferiors, tuning this off treats the kernel as a single inferior with many threads. Useful to avoid bugs in current gdb Command line option to be added later

Readd the different base address for gdb

4ef1fec

Currently it appears impossible to relocate the current executable in gdb https://sourceware.org/bugzilla/show_bug.cgi?id=20007 Also disable multiprocess mode by default, since it breaks upstream gdb

ariscop force-pushed the kdgdb branch from 892928e to 4ef1fec Compare May 8, 2016 09:41

Storing working dir changes, don't plan to work on this anymore

59ad4fd

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Kdgdb Changes #12

Kdgdb Changes #12

ariscop commented Apr 20, 2016

gigaherz commented Apr 20, 2016

learn-more Apr 20, 2016

ariscop Apr 20, 2016

ariscop commented Apr 23, 2016

ariscop commented Apr 23, 2016

ariscop commented Apr 28, 2016

ariscop commented Apr 29, 2016

ariscop commented May 9, 2016

ariscop commented Jun 6, 2016

AmineKhaldi commented Sep 14, 2016

Kdgdb Changes #12

Are you sure you want to change the base?

Kdgdb Changes #12

Conversation

ariscop commented Apr 20, 2016

gigaherz commented Apr 20, 2016

learn-more Apr 20, 2016

Choose a reason for hiding this comment

ariscop Apr 20, 2016

Choose a reason for hiding this comment

ariscop commented Apr 23, 2016

ariscop commented Apr 23, 2016

ariscop commented Apr 28, 2016

ariscop commented Apr 29, 2016

ariscop commented May 9, 2016

ariscop commented Jun 6, 2016

AmineKhaldi commented Sep 14, 2016