[lldb-dev] Fwd: [Lldb-commits] [lldb] r216238 - Add missing HostInfo::Initialize() in llgs.

Tue Aug 26 15:03:16 PDT 2014

I'm picking my battles in the interest of getting things done, Zach :-P

On Tue, Aug 26, 2014 at 3:02 PM, Zachary Turner <zturner at google.com> wrote:

> I feel like relying on dead code elimination is a crutch to workaround an
> improperly layered set of libraries.  It gets the job done with the least
> amount of programmer effort involved, but it might be worth at least
> brainstorming a better long-term solution and finding a way to split the
> libraries apart in such a way that llgs can bring in only the things it
> needs.  Even if doing the actual split up front might be untenable due to
> time constraints or other factors, having a clear direction regarding an
> "end state" would perhaps allow new code to get fit into this model when
> it's written, and refactoring toward the end goal to happen gradually over
> time, as time permits.
>
>
> On Tue, Aug 26, 2014 at 12:21 PM, Todd Fiala <tfiala at google.com> wrote:
>
>> So here I *think* we want to do something that will let the linker dead
>> code elimination be able to figure out which modules it doesn't need.
>>  Which means I think we really want the call graph for initializing the
>> lighter apps (lldb-platform, lldb-gdbserver) to have a method they call
>> which initializes the always-present items (i.e. the core).  Then we can
>> have the heavier apps (lldb) call the core initialization, then layer on
>> the higher-level systems.
>>
>>
>>
>> ---------- Forwarded message ----------
>> From: Todd Fiala <tfiala at google.com>
>> Date: Tue, Aug 26, 2014 at 12:18 PM
>> Subject: Re: [Lldb-commits] [lldb] r216238 - Add missing
>> HostInfo::Initialize() in llgs.
>> To: Greg Clayton <gclayton at apple.com>
>> Cc: Todd Fiala <todd.fiala at gmail.com>, lldb-commits <
>> lldb-commits at cs.uiuc.edu>
>>
>>
>> Ok back looking at this for real.
>>
>> So I'm going to need to adjust Debugger::Initalize() or use an
>> alternative installation path.  Right now it goes and calls
>> lldb_private::Initialize(), which pretty much brings in the whole world.
>>  llgs needs to be *thin*, as we're counting on static linkage and dead code
>> elimination to cut out the bulk of the exe size.  There is a ton more stuff
>> initialized in lldb_private::Initialize() than we need in llgs.
>>
>> (e.g. ProcessLinux, DynamicLoader*, language runtimes, etc. --- none of
>> those should be getting initialized in llgs).
>>
>>
>>
>>
>> On Tue, Aug 26, 2014 at 11:52 AM, Todd Fiala <tfiala at google.com> wrote:
>>
>>> Ok - but - are we okay with moving PluginManager::Initialize() within
>>> there?  If so, we're good.  Because HostInfo::Initialize() needs to come
>>> *before* PluginManager::Initialize().
>>>
>>> Seems reasonable, I just need to peek where we're initializing
>>> PluginManager.
>>>
>>>
>>> On Tue, Aug 26, 2014 at 11:51 AM, Greg Clayton <gclayton at apple.com>
>>> wrote:
>>>
>>>> then HostInfo::Initialize() should go first within
>>>> Debugger::Initialize() and HostInfo::Terminate() should go last in
>>>> Debugger::Terminate().
>>>>
>>>> > On Aug 26, 2014, at 11:46 AM, Todd Fiala <tfiala at google.com> wrote:
>>>> >
>>>> > > But I think Debugger depends on HostInfo being initialized first.
>>>> >
>>>> > Rather I meant:
>>>> > * PluginManager::Initialize() requires HostInfo to already be
>>>> initialized.
>>>> > * Debugger::Initialize() expects PluginManager is already initialized.
>>>> >
>>>> >
>>>> > On Tue, Aug 26, 2014 at 11:44 AM, Todd Fiala <tfiala at google.com>
>>>> wrote:
>>>> > There might be an issue there.
>>>> >
>>>> > PluginManager uses the file system initialization.  But I think
>>>> Debugger depends on HostInfo being initialized first.
>>>> >
>>>> > I'll look and see if I can detangle that.
>>>> >
>>>> >
>>>> > On Tue, Aug 26, 2014 at 11:43 AM, Todd Fiala <tfiala at google.com>
>>>> wrote:
>>>> > Ah ok.  I'll move those in a minute.  Another check-in coming shortly
>>>> before.
>>>> >
>>>> >
>>>> > On Tue, Aug 26, 2014 at 11:34 AM, Greg Clayton <gclayton at apple.com>
>>>> wrote:
>>>> > Todd:
>>>> >
>>>> > HostInfo::Intialize() should be done in Debugger::Initialize()
>>>> > HostInfo::Terminate() should be done in Debugger::Terminate()
>>>> >
>>>> > We want there to be single call to initialize everything and
>>>> terminate everything and this is done via the Debugger calls.
>>>> >
>>>> > > On Aug 21, 2014, at 3:41 PM, Todd Fiala <todd.fiala at gmail.com>
>>>> wrote:
>>>> > >
>>>> > > Author: tfiala
>>>> > > Date: Thu Aug 21 17:41:25 2014
>>>> > > New Revision: 216238
>>>> > >
>>>> > > URL: http://llvm.org/viewvc/llvm-project?rev=216238&view=rev
>>>> > > Log:
>>>> > > Add missing HostInfo::Initialize() in llgs.
>>>> > >
>>>> > > This fixes an llgs segfault on startup.
>>>> > >
>>>> > > Modified:
>>>> > >    lldb/trunk/tools/lldb-gdbserver/lldb-gdbserver.cpp
>>>> > >
>>>> > > Modified: lldb/trunk/tools/lldb-gdbserver/lldb-gdbserver.cpp
>>>> > > URL:
>>>> http://llvm.org/viewvc/llvm-project/lldb/trunk/tools/lldb-gdbserver/lldb-gdbserver.cpp?rev=216238&r1=216237&r2=216238&view=diff
>>>> > >
>>>> ==============================================================================
>>>> > > --- lldb/trunk/tools/lldb-gdbserver/lldb-gdbserver.cpp (original)
>>>> > > +++ lldb/trunk/tools/lldb-gdbserver/lldb-gdbserver.cpp Thu Aug 21
>>>> 17:41:25 2014
>>>> > > @@ -32,6 +32,7 @@
>>>> > > #include "lldb/Core/Debugger.h"
>>>> > > #include "lldb/Core/PluginManager.h"
>>>> > > #include "lldb/Core/StreamFile.h"
>>>> > > +#include "lldb/Host/HostInfo.h"
>>>> > > #include "lldb/Host/OptionParser.h"
>>>> > > #include "lldb/Host/Socket.h"
>>>> > > #include "lldb/Interpreter/CommandInterpreter.h"
>>>> > > @@ -152,6 +153,7 @@ dump_available_platforms (FILE *output_f
>>>> > > static void
>>>> > > initialize_lldb_gdbserver ()
>>>> > > {
>>>> > > +    HostInfo::Initialize ();
>>>> > >     PluginManager::Initialize ();
>>>> > >     Debugger::Initialize (NULL);
>>>> > > }
>>>> > > @@ -159,7 +161,7 @@ initialize_lldb_gdbserver ()
>>>> > > static void
>>>> > > terminate_lldb_gdbserver ()
>>>> > > {
>>>> > > -    Debugger::Terminate();
>>>> > > +    Debugger::Terminate ();
>>>> > >     PluginManager::Terminate ();
>>>> > > }
>>>> > >
>>>> > >
>>>> > >
>>>> > > _______________________________________________
>>>> > > lldb-commits mailing list
>>>> > > lldb-commits at cs.uiuc.edu
>>>> > > http://lists.cs.uiuc.edu/mailman/listinfo/lldb-commits
>>>> >
>>>> > _______________________________________________
>>>> > lldb-commits mailing list
>>>> > lldb-commits at cs.uiuc.edu
>>>> > http://lists.cs.uiuc.edu/mailman/listinfo/lldb-commits
>>>> >
>>>> >
>>>> >
>>>> > --
>>>> > Todd Fiala |   Software Engineer |     tfiala at google.com |
>>>> 650-943-3180
>>>> >
>>>> >
>>>> >
>>>> >
>>>> > --
>>>> > Todd Fiala |   Software Engineer |     tfiala at google.com |
>>>> 650-943-3180
>>>> >
>>>> >
>>>> >
>>>> >
>>>> > --
>>>> > Todd Fiala |   Software Engineer |     tfiala at google.com |
>>>> 650-943-3180
>>>> >
>>>>
>>>>
>>>
>>>
>>> --
>>> Todd Fiala | Software Engineer |  tfiala at google.com |  650-943-3180
>>>
>>
>>
>>
>> --
>> Todd Fiala | Software Engineer |  tfiala at google.com |  650-943-3180
>>
>>
>>
>> --
>> Todd Fiala | Software Engineer |  tfiala at google.com |  650-943-3180
>>
>> _______________________________________________
>> lldb-dev mailing list
>> lldb-dev at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/lldb-dev
>>
>>
>

-- 
Todd Fiala | Software Engineer | tfiala at google.com | 650-943-3180
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/lldb-dev/attachments/20140826/e3bccb3f/attachment.html>