Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Sun, Dec 6, 2009 at 2:51 PM, Øyvind Harboe oyvind.har...@zylin.com wrote: 1. Move the $(srcdir) from defining _DIR vars to their use. i.e. set just the directory component. I'm not quite sure what you mean by this. I have a makefile rule that is only used with the minidriver, would you like that to be defined always? I'm not even sure that would work when there are definitions missing... The oharboe/fixminidriver code is shaping up, but I still need to run tests. oharboe/fixminidriver now builds with: Linux: a) --enable-dummy b) --enable-minidriver-dummy c) zy1000 Cygwin: a) --enable-dummy b) --enable-minidriver-dummy Some problems: make clean does not remove jtag_minidriver.h and minidriver_impl.h. -- Øyvind Harboe US toll free 1-866-980-3434 / International +47 51 63 25 00 http://www.zylin.com/zy1000.html ARM7 ARM9 ARM11 XScale Cortex JTAG debugger and flash programmer ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Mon, 2009-12-07 at 12:26 +0100, Øyvind Harboe wrote: On Sun, Dec 6, 2009 at 2:51 PM, Øyvind Harboe oyvind.har...@zylin.com wrote: 1. Move the $(srcdir) from defining _DIR vars to their use. i.e. set just the directory component. I'm not quite sure what you mean by this. I have a makefile rule that is only used with the minidriver, would you like that to be defined always? I'm not even sure that would work when there are definitions missing... The oharboe/fixminidriver code is shaping up, but I still need to run tests. oharboe/fixminidriver now builds with: Linux: a) --enable-dummy b) --enable-minidriver-dummy c) zy1000 Cygwin: a) --enable-dummy b) --enable-minidriver-dummy Some problems: make clean does not remove jtag_minidriver.h and minidriver_impl.h. And if you read _all_ of the instructions I gave you, this would not be a problem. When you think they are done, post these patches in a new thread. --Z ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
1. Move the $(srcdir) from defining _DIR vars to their use. i.e. set just the directory component. I'm not quite sure what you mean by this. I have a makefile rule that is only used with the minidriver, would you like that to be defined always? I'm not even sure that would work when there are definitions missing... The oharboe/fixminidriver code is shaping up, but I still need to run tests. -- Øyvind Harboe US toll free 1-866-980-3434 / International +47 51 63 25 00 http://www.zylin.com/zy1000.html ARM7 ARM9 ARM11 XScale Cortex JTAG debugger and flash programmer ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Sat, 2009-12-05 at 08:25 +0100, Øyvind Harboe wrote: This worked pretty well, but there are lots of (to me) inexplicable -I's on the command line still and none of them point to the right place in the build tree. I'm testing minidummy build != src dir. oharboe/fixminidriver2 ../openocd/configure --enable-minidriver-dummy --enable-maintainer-mode make libtool: compile: gcc -std=gnu99 -DHAVE_CONFIG_H -I. -I../../../zy1000/openocd/src/jtag -I../.. -I../../../zy1000/openocd/src -I../../../zy1000/openocd/src/jtag/minidriver -g -O2 -Wall -Wstrict-prototypes -Wformat-security -Wextra -Wno-unused-parameter -Wbad-function-cast -Wcast-align -Wredundant-decls -Werror -MT core.lo -MD -MP -MF .deps/core.Tpo -c ../../../zy1000/openocd/src/jtag/core.c -o core.o In file included from ../../../zy1000/openocd/src/jtag/jtag.h:729, from ../../../zy1000/openocd/src/jtag/core.c:34: ../../../zy1000/openocd/src/jtag/minidriver.h:50:33: error: jtag/minidriver_imp.h: No such file or directory cc1: warnings being treated as errors At the least, the -I$(srcdir)/jtag/minidriver should be gone, and I'd guess you didn't add the two header names to BUILT_SOURCES so their rules get called before everything else. At this point, I need to see a new series, so please mirror or post them. I don't know what you've done to tell you how to fix it. :) --Z ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Sat, Dec 5, 2009 at 8:45 AM, Zach Welch z...@superlucidity.net wrote: On Sat, 2009-12-05 at 08:35 +0100, Øyvind Harboe wrote: The approach seems to work fine, but requires that some other issues are sorted out first with include dirs. From error message above, there are two strange -I additions: NB! build != src dir -I. = current build directory -I../.. = this is the root build dir, so I would have to include include src/jtag/minidriver_imp.h. Was this supposed to be the build/src dir instead? I don't know where the two above are defined, but at the top of each Makefile.am there is a AM_CPPFLAGS = \ -I$(top_srcdir)/src Should this include the build/src dir as well? You got it. We need to add -I$(top_builddir)/src to all directories now that we are talking about generating them on the fly. Do that in a separate patch please. :) Seems to build now (oharboe/fixminidriver3) I need to clean up and run tests, but it should be plain sailing from here on. -- Øyvind Harboe US toll free 1-866-980-3434 / International +47 51 63 25 00 http://www.zylin.com/zy1000.html ARM7 ARM9 ARM11 XScale Cortex JTAG debugger and flash programmer ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Sat, 2009-12-05 at 11:14 +0100, Øyvind Harboe wrote: On Sat, Dec 5, 2009 at 8:45 AM, Zach Welch z...@superlucidity.net wrote: On Sat, 2009-12-05 at 08:35 +0100, Øyvind Harboe wrote: The approach seems to work fine, but requires that some other issues are sorted out first with include dirs. From error message above, there are two strange -I additions: NB! build != src dir -I. = current build directory -I../.. = this is the root build dir, so I would have to include include src/jtag/minidriver_imp.h. Was this supposed to be the build/src dir instead? I don't know where the two above are defined, but at the top of each Makefile.am there is a AM_CPPFLAGS = \ -I$(top_srcdir)/src Should this include the build/src dir as well? You got it. We need to add -I$(top_builddir)/src to all directories now that we are talking about generating them on the fly. Do that in a separate patch please. :) Seems to build now (oharboe/fixminidriver3) I need to clean up and run tests, but it should be plain sailing from here on. Patch #1: 1. use continuations '\' to make the files more patch-friendly. Patch #3: 1. Move the $(srcdir) from defining _DIR vars to their use. i.e. set just the directory component. 2. Set BUILD_SOURCES as empty and use += in both branches for symmetry. 3. ZY1000_POKE does not need to be here (and #if 0?!? never!!!) Cheers, --Z ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
It's getting there, see oharboe/fixminidriver. I had to mess with lots of Makefile.am to pollute the include path. If I can get that out of the way, then I should be able to clean up the rest too. -- Øyvind Harboe US toll free 1-866-980-3434 / International +47 51 63 25 00 http://www.zylin.com/zy1000.html ARM7 ARM9 ARM11 XScale Cortex JTAG debugger and flash programmer ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Fri, 2009-12-04 at 16:03 +0100, Øyvind Harboe wrote: It's getting there, see oharboe/fixminidriver. I had to mess with lots of Makefile.am to pollute the include path. If I can get that out of the way, then I should be able to clean up the rest too Yeah, those changes should not go into the tree as-is. They make a total mess of the build system. In your second patch, the drivers/minidriver_imp.h patch is close, but a little buggy. You #define and declare the same function; the declared version should be the _imp version. Seeing your new patch gives me a new idea for how to fix things: 1) In jtag/Makefile, use BUILT_SOURCES to copy the proper minidriver_imp.h (and jtag_minidriver.h, if needed) into the jtag directory. 2) #include jtag/minidriver_imp.h and jtag/jtag_minidriver_h. Will that work? It should, allowing to kill your first patch too. --Z ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Fri, Dec 4, 2009 at 10:36 PM, Zach Welch z...@superlucidity.net wrote: On Fri, 2009-12-04 at 16:03 +0100, Øyvind Harboe wrote: It's getting there, see oharboe/fixminidriver. I had to mess with lots of Makefile.am to pollute the include path. If I can get that out of the way, then I should be able to clean up the rest too Yeah, those changes should not go into the tree as-is. They make a total mess of the build system. Yeah, I had to put in a kludge in the whole 15 minutes I had between routing FPGAs to test the rest :-) In your second patch, the drivers/minidriver_imp.h patch is close, but a little buggy. You #define and declare the same function; the declared version should be the _imp version. I think I can simply do away with the _imp and the #define altogether. More straightforward. Seeing your new patch gives me a new idea for how to fix things: 1) In jtag/Makefile, use BUILT_SOURCES to copy the proper minidriver_imp.h (and jtag_minidriver.h, if needed) into the jtag directory. Sounds good to me. I don't know the autotools speak to do that. I'll give it a go. 2) #include jtag/minidriver_imp.h and jtag/jtag_minidriver_h. minidriver_imp.h includes jtag_minidriver.h. Will that work? It should, allowing to kill your first patch too. Should do. Some trivial Makefile.am is the biggest hurdle for me, my autotools skills are pretty limited. -- Øyvind Harboe US toll free 1-866-980-3434 / International +47 51 63 25 00 http://www.zylin.com/zy1000.html ARM7 ARM9 ARM11 XScale Cortex JTAG debugger and flash programmer ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Fri, 2009-12-04 at 22:51 +0100, Øyvind Harboe wrote: On Fri, Dec 4, 2009 at 10:36 PM, Zach Welch z...@superlucidity.net wrote: On Fri, 2009-12-04 at 16:03 +0100, Øyvind Harboe wrote: It's getting there, see oharboe/fixminidriver. I had to mess with lots of Makefile.am to pollute the include path. If I can get that out of the way, then I should be able to clean up the rest too Yeah, those changes should not go into the tree as-is. They make a total mess of the build system. Yeah, I had to put in a kludge in the whole 15 minutes I had between routing FPGAs to test the rest :-) In your second patch, the drivers/minidriver_imp.h patch is close, but a little buggy. You #define and declare the same function; the declared version should be the _imp version. I think I can simply do away with the _imp and the #define altogether. More straightforward. Minimal patches. Whatever that turns out to be. :) Seeing your new patch gives me a new idea for how to fix things: 1) In jtag/Makefile, use BUILT_SOURCES to copy the proper minidriver_imp.h (and jtag_minidriver.h, if needed) into the jtag directory. Sounds good to me. I don't know the autotools speak to do that. I'll give it a go. I would have to stab at it several times myself, I'm sure. At the core, I would expect the rules to look something like: BUILT_SOURCES += minidriver_imp.h jtag_minidriver.h minidriver_imp.h: $(MINIDRIVER_IMP_DIR)/minidriver_imp.h cp $ $@ jtag_minidriver.h: $(JTAG_MINIDRIVER_DIR)/jtag_minidriver.h cp $ $@ The two _DIR variables get set inside the automake logic. The values will be 'drivers' and 'minidriver' for the first, and 'minidummy' and 'zy1000' for the second. So, you're just copying the two files from the source directories (rather than -I'ing those directories). You'll also need to add those files to the CLEANFILES rule. 2) #include jtag/minidriver_imp.h and jtag/jtag_minidriver_h. minidriver_imp.h includes jtag_minidriver.h. Right, but both should use bracketed paths... not foo.h... This should obviate the rules added by your first patch. --Z ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
This worked pretty well, but there are lots of (to me) inexplicable -I's on the command line still and none of them point to the right place in the build tree. I'm testing minidummy build != src dir. oharboe/fixminidriver2 ../openocd/configure --enable-minidriver-dummy --enable-maintainer-mode make libtool: compile: gcc -std=gnu99 -DHAVE_CONFIG_H -I. -I../../../zy1000/openocd/src/jtag -I../.. -I../../../zy1000/openocd/src -I../../../zy1000/openocd/src/jtag/minidriver -g -O2 -Wall -Wstrict-prototypes -Wformat-security -Wextra -Wno-unused-parameter -Wbad-function-cast -Wcast-align -Wredundant-decls -Werror -MT core.lo -MD -MP -MF .deps/core.Tpo -c ../../../zy1000/openocd/src/jtag/core.c -o core.o In file included from ../../../zy1000/openocd/src/jtag/jtag.h:729, from ../../../zy1000/openocd/src/jtag/core.c:34: ../../../zy1000/openocd/src/jtag/minidriver.h:50:33: error: jtag/minidriver_imp.h: No such file or directory cc1: warnings being treated as errors -- Øyvind Harboe US toll free 1-866-980-3434 / International +47 51 63 25 00 http://www.zylin.com/zy1000.html ARM7 ARM9 ARM11 XScale Cortex JTAG debugger and flash programmer ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Sat, 2009-12-05 at 08:35 +0100, Øyvind Harboe wrote: The approach seems to work fine, but requires that some other issues are sorted out first with include dirs. From error message above, there are two strange -I additions: NB! build != src dir -I. = current build directory -I../.. = this is the root build dir, so I would have to include include src/jtag/minidriver_imp.h. Was this supposed to be the build/src dir instead? I don't know where the two above are defined, but at the top of each Makefile.am there is a AM_CPPFLAGS = \ -I$(top_srcdir)/src Should this include the build/src dir as well? You got it. We need to add -I$(top_builddir)/src to all directories now that we are talking about generating them on the fly. Do that in a separate patch please. :) --Z ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
So the minidriver (including classic API) must have control over whether or
not the following fn's gets inlined or not. These fn's really are in the inner
loops and boil away to *nothing* on a low performance low latency
system.
Note that I'm not keen on the idea of an ABI for OpenOCD, we need
more flexibility and less commitment than that, but I think the goal
can be a good design influence.
static inline void jtag_add_dr_out(struct jtag_tap* tap,
int num_fields, const int* num_bits, const uint32_t* value,
tap_state_t end_state)
{
assert(end_state != TAP_RESET);
assert(end_state != TAP_INVALID);
cmd_queue_cur_state = end_state;
interface_jtag_add_dr_out(tap,
num_fields, num_bits, value,
end_state);
}
/* The minidriver must be able to inline this fn */
static inline void jtag_add_callback(jtag_callback1_t f,
jtag_callback_data_t data0)
{
interface_jtag_add_callback(f, data0);
}
/* The minidriver must be able to inline this fn */
static inline void jtag_add_callback4(jtag_callback_t f,
jtag_callback_data_t data0,
jtag_callback_data_t data1, jtag_callback_data_t data2,
jtag_callback_data_t data3)
{
interface_jtag_add_callback4(f, data0, data1, data2, data3);
}
--
Øyvind Harboe
US toll free 1-866-980-3434 / International +47 51 63 25 00
http://www.zylin.com/zy1000.html
ARM7 ARM9 ARM11 XScale Cortex
JTAG debugger and flash programmer
___
Openocd-development mailing list
Openocd-development@lists.berlios.de
https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Thu, 2009-12-03 at 09:20 +0100, Øyvind Harboe wrote: The public API _must_ be a non-inline version of the API that calls the inline version. Period. You don't explain why this is necessary and I don't agree. I believe you assume that it is impossible to allow the minidriver to inline *and* support an ABI. Actually you can, but the choice is made compile time on the particular platform. The minidriver should be able to take over and *inline* jtag_add_xxx() fn's if that is the right thing to do on a particular platform. This is an important point: if you want your ZY1000 to provide an inline version, then you should be able to do that. Similarly the classic minidriver must be able to implement jtag_add_xxx() as non-inline fn's to support an ABI. Inline functions may not be used in public APIs, because they prevent upgrading shared libraries. The statement above is wrong. For a platform with an ABI for shared libraries you use a minidriver(the classic one) that *does* allow shared library ABI, on an embedded platform you use the minidriver that removes this unecessary layer. Right, for situations like yours, it does makes sense to allow inline functions. For shared libraries, however, my statements stand. And those are public APIs. Your platform has no public APIs and never will, because you're statically linking the everything into one binary... right? So my statements are fully correct, and your reading of them was too incorrect. ;) Anyway, I think that I might have fixed everything that I was annoyed with in the patches I just pushed. Take another look at things and tell me what you think needs to be done now :) --Z ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Thu, 2009-12-03 at 09:53 +0100, Øyvind Harboe wrote: So the minidriver (including classic API) must have control over whether or not the following fn's gets inlined or not. These fn's really are in the inner loops and boil away to *nothing* on a low performance low latency system. And they should now, after my patches. I want to see yours again, updated against the current master. :) Note that I'm not keen on the idea of an ABI for OpenOCD, we need more flexibility and less commitment than that, but I think the goal can be a good design influence. For your platform, you can forget about the ABI. For shared libraries, it is important However, I agree that commitment to the ABI should not be something to consider until 1.0 (or even 2.0...). Striving to create a stable ABI, however, will improve the code, and that is the real goal here. It's about architectural correctness, and inline functions are _not_ necessary for good performance. I feel very confident that your requirement for them could be eliminated by using better algorithms in the JTAG layer, and you would still see better performance than you get today. Sure, that entails far more work, but the price to get correctness in design is higher than that required to reach correctness in function. --Z ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Thu, 2009-12-03 at 10:22 +0100, Øyvind Harboe wrote: (Ref. discussion I think we've settled that the minidriver can either implement an ABI or use inlining for performance.) On Thu, Dec 3, 2009 at 10:11 AM, Zach Welch z...@superlucidity.net wrote: On Thu, 2009-12-03 at 09:53 +0100, Øyvind Harboe wrote: So the minidriver (including classic API) must have control over whether or not the following fn's gets inlined or not. These fn's really are in the inner loops and boil away to *nothing* on a low performance low latency system. And they should now, after my patches. I want to see yours again, updated against the current master. :) I have an automake problem / include file problem I don't know how to solve in a pretty fashion: jtag.h needs to include a minidriver_inline.h for the active minidriver. How to do that? You should not need to add any new files at this point: jtag/minidriver.h includes minidriver_imp.h: jtag/drivers/minidriver_imp.h -- the driver's implementation jtag/minidriver/minidriver_imp.h -- the minidriver's implementation The minidriver's implementaiton includes jtag_minidriver.h, which is defined by the implementation: jtag/minidummy/jtag_minidriver.h jtag/zy1000/jtag_minidriver.h Put stuff in these files... Do not put new inline functions in jtag/jtag.h. jtag/minidriver.h, or jtag/drivers/minidriver_imp.h. - I don't want to add -I to *all* places that #include jtag.h to the active minidriver - #include ACTIVE_MINIDRIVER, using a #define? I don't particularly like the #include jtag.h form. Why quotes rather than angle brackets BTW? Quotes allow including from the current directory, why should we allow the official jtag.h to be replaced by a variant local to a source dir? Can't think of why really. Look at my mirror. Go there right now. :) I've mentioned my 'hdr' and 'tests' branches a few times. You haven't looked at them, or you wouldn't be asking this question. That work is now on my master branch, queued to be pushed. Indeed, I was just about to reply to my post on this subject (see my recent post with the subject libopenocd... again?) to warn that I want to push this series in the next couple of hours. So, you've gone and read it now, right... did it answer your question? Unfortunately, it needs a little more love before it's ready to push, but the above minidriver_imp.h and jtag_minidriver.h will _not_ be changed to use angle brackets... Otherwise, these tricks don't work. --Z ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Thu, Dec 3, 2009 at 10:36 AM, Zach Welch z...@superlucidity.net wrote: On Thu, 2009-12-03 at 10:22 +0100, Øyvind Harboe wrote: (Ref. discussion I think we've settled that the minidriver can either implement an ABI or use inlining for performance.) On Thu, Dec 3, 2009 at 10:11 AM, Zach Welch z...@superlucidity.net wrote: On Thu, 2009-12-03 at 09:53 +0100, Øyvind Harboe wrote: So the minidriver (including classic API) must have control over whether or not the following fn's gets inlined or not. These fn's really are in the inner loops and boil away to *nothing* on a low performance low latency system. And they should now, after my patches. I want to see yours again, updated against the current master. :) I have an automake problem / include file problem I don't know how to solve in a pretty fashion: jtag.h needs to include a minidriver_inline.h for the active minidriver. How to do that? You should not need to add any new files at this point: jtag/minidriver.h includes minidriver_imp.h: jtag/drivers/minidriver_imp.h -- the driver's implementation jtag/minidriver/minidriver_imp.h -- the minidriver's implementation The minidriver's implementaiton includes jtag_minidriver.h, which is defined by the implementation: jtag/minidummy/jtag_minidriver.h jtag/zy1000/jtag_minidriver.h Put stuff in these files... Do not put new inline functions in jtag/jtag.h. jtag/minidriver.h, or jtag/drivers/minidriver_imp.h. To get inlining to work, I need to make a few fn's in core.c inline fn's too. Take jtag_add_dr_out() as an example. jtag_add_dr_out() must be defined after including jtag.h. This means jtag.h must include the inline version of interface_jtag_add_dr_out(). So yes... I need to figure out a way for minidrivers to expose, via jtag.h, the inline implementations. The driver minidriver should be able to implement jtag_add_dr_out() as a non-inlined function in order to be able to support an ABI at that level. Look at my mirror. Go there right now. :) I've mentioned my 'hdr' and 'tests' branches a few times. You haven't looked at them, or you wouldn't be asking this question. That work is now on my master branch, queued to be pushed. Rights... I recall those changes. Yes that's a good cleanup. I protest when you are doing cleanup work that causes problems, but not comment when it looks all good. I think I want to wait with the minidriver work until you have the hdr branch pushed. -- Øyvind Harboe US toll free 1-866-980-3434 / International +47 51 63 25 00 http://www.zylin.com/zy1000.html ARM7 ARM9 ARM11 XScale Cortex JTAG debugger and flash programmer ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
(Ref. discussion I think we've settled that the minidriver can either implement an ABI or use inlining for performance.) On Thu, Dec 3, 2009 at 10:11 AM, Zach Welch z...@superlucidity.net wrote: On Thu, 2009-12-03 at 09:53 +0100, Øyvind Harboe wrote: So the minidriver (including classic API) must have control over whether or not the following fn's gets inlined or not. These fn's really are in the inner loops and boil away to *nothing* on a low performance low latency system. And they should now, after my patches. I want to see yours again, updated against the current master. :) I have an automake problem / include file problem I don't know how to solve in a pretty fashion: jtag.h needs to include a minidriver_inline.h for the active minidriver. How to do that? - I don't want to add -I to *all* places that #include jtag.h to the active minidriver - #include ACTIVE_MINIDRIVER, using a #define? I don't particularly like the #include jtag.h form. Why quotes rather than angle brackets BTW? Quotes allow including from the current directory, why should we allow the official jtag.h to be replaced by a variant local to a source dir? Can't think of why really. -- Øyvind Harboe US toll free 1-866-980-3434 / International +47 51 63 25 00 http://www.zylin.com/zy1000.html ARM7 ARM9 ARM11 XScale Cortex JTAG debugger and flash programmer ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Thu, 2009-12-03 at 10:51 +0100, Øyvind Harboe wrote: On Thu, Dec 3, 2009 at 10:36 AM, Zach Welch z...@superlucidity.net wrote: On Thu, 2009-12-03 at 10:22 +0100, Øyvind Harboe wrote: (Ref. discussion I think we've settled that the minidriver can either implement an ABI or use inlining for performance.) On Thu, Dec 3, 2009 at 10:11 AM, Zach Welch z...@superlucidity.net wrote: On Thu, 2009-12-03 at 09:53 +0100, Øyvind Harboe wrote: So the minidriver (including classic API) must have control over whether or not the following fn's gets inlined or not. These fn's really are in the inner loops and boil away to *nothing* on a low performance low latency system. And they should now, after my patches. I want to see yours again, updated against the current master. :) I have an automake problem / include file problem I don't know how to solve in a pretty fashion: jtag.h needs to include a minidriver_inline.h for the active minidriver. How to do that? You should not need to add any new files at this point: jtag/minidriver.h includes minidriver_imp.h: jtag/drivers/minidriver_imp.h -- the driver's implementation jtag/minidriver/minidriver_imp.h -- the minidriver's implementation The minidriver's implementaiton includes jtag_minidriver.h, which is defined by the implementation: jtag/minidummy/jtag_minidriver.h jtag/zy1000/jtag_minidriver.h Put stuff in these files... Do not put new inline functions in jtag/jtag.h. jtag/minidriver.h, or jtag/drivers/minidriver_imp.h. To get inlining to work, I need to make a few fn's in core.c inline fn's too. Take jtag_add_dr_out() as an example. Well, you can't. The rules prohibit this. ;) jtag_add_dr_out() must be defined after including jtag.h. This means jtag.h must include the inline version of interface_jtag_add_dr_out(). That's fine -- for ZY1000. So yes... I need to figure out a way for minidrivers to expose, via jtag.h, the inline implementations. ...while also providing an out-of-line version for the drivers. :/ Define a macro called jtag_add_dr_out in drivers/minidriver_imp.h that redirects the call to an out-of-line function (with the _imp suffix). The driver.c file can define that function and #include a new header that is also shared with minidriver/minidriver_imp.h. This new header provides the inline version, which is called from the driver's _imp. There you have it. Or, improve the algorithms such that such inlining is moot. ;) The driver minidriver should be able to implement jtag_add_dr_out() as a non-inlined function in order to be able to support an ABI at that level. As I said elsewhere, it's now safe to #include minidriver.h in jtag.h, so long as you do not add in-lines to the files I noted. Look at my mirror. Go there right now. :) I've mentioned my 'hdr' and 'tests' branches a few times. You haven't looked at them, or you wouldn't be asking this question. That work is now on my master branch, queued to be pushed. Rights... I recall those changes. Yes that's a good cleanup. I protest when you are doing cleanup work that causes problems, but not comment when it looks all good. The problem is that your definition of good puts performance too high at the list of priorities. At this stage, that is like polishing a turd. You would be better off investigating long-term fixes, rather than chasing around implementation issues that patch over the problem temporary. The architectural improvements that are binding your hands presently reflect other design improvements need to be made. That said, I still want to see you find a suitable workaround, and there will undoubtedly be a compromise involved. ;) I think I want to wait with the minidriver work until you have the hdr branch pushed. It's nearly ready. --Z ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
Do not apply this until you fix the problem noted below.
Thanks,
Zach
On Wed, 2009-12-02 at 14:38 +0100, Øyvind Harboe wrote:
Fix 'regression' where jtag_add_dr_out() is no longer inlined
after refactoring.
The minidrivers strip away all the overhead and allow inner loops
to communicate directly with a hardware FIFO.
Signed-off-by: Øyvind Harboe oyvind.har...@zylin.com
---
src/jtag/Makefile.am |2 -
src/jtag/core.c | 14 -
src/jtag/driver.c| 10 +++
src/jtag/jtag.h | 37 +++--
src/jtag/minidriver.h| 67 ++---
src/jtag/zy1000/zy1000.c |3 +-
6 files changed, 73 insertions(+), 60 deletions(-)
diff --git a/src/jtag/Makefile.am b/src/jtag/Makefile.am
index 5254a2b..c762300 100644
--- a/src/jtag/Makefile.am
+++ b/src/jtag/Makefile.am
@@ -11,11 +11,9 @@ if MINIDRIVER
if ZY1000
DRIVERFILES += zy1000/zy1000.c
-AM_CPPFLAGS += -I$(srcdir)/zy1000
endif
if MINIDRIVER_DUMMY
DRIVERFILES += minidummy/minidummy.c commands.c
-AM_CPPFLAGS += -I$(srcdir)/minidummy
endif
else
diff --git a/src/jtag/core.c b/src/jtag/core.c
index 9230cc2..cf14394 100644
--- a/src/jtag/core.c
+++ b/src/jtag/core.c
@@ -502,20 +502,6 @@ void jtag_add_plain_dr_scan(int in_num_fields, const
struct scan_field *in_field
jtag_set_error(retval);
}
-void jtag_add_dr_out(struct jtag_tap* tap,
- int num_fields, const int* num_bits, const uint32_t* value,
- tap_state_t end_state)
-{
- assert(end_state != TAP_RESET);
- assert(end_state != TAP_INVALID);
-
- cmd_queue_cur_state = end_state;
-
- interface_jtag_add_dr_out(tap,
- num_fields, num_bits, value,
- end_state);
-}
-
void jtag_add_tlr(void)
{
jtag_prelude(TAP_RESET);
diff --git a/src/jtag/driver.c b/src/jtag/driver.c
index cadd88e..fe56369 100644
--- a/src/jtag/driver.c
+++ b/src/jtag/driver.c
@@ -525,3 +525,13 @@ void interface_jtag_add_callback(jtag_callback1_t
callback, jtag_callback_data_t
jtag_add_callback4(jtag_convert_to_callback4, data0,
(jtag_callback_data_t)callback, 0, 0);
}
+void interface_jtag_alloc_in_value32(struct scan_field *field)
+{
+ field-in_value = (uint8_t *)cmd_queue_alloc(4);
+}
+
+void interface_jtag_add_scan_check_alloc(struct scan_field *field)
+{
+ unsigned num_bytes = DIV_ROUND_UP(field-num_bits, 8);
+ field-in_value = (uint8_t *)cmd_queue_alloc(num_bytes);
+}
diff --git a/src/jtag/jtag.h b/src/jtag/jtag.h
index ee96775..4b0e8b2 100644
--- a/src/jtag/jtag.h
+++ b/src/jtag/jtag.h
@@ -2,7 +2,7 @@
* Copyright (C) 2005 by Dominic Rath*
* dominic.r...@gmx.de *
* *
-* Copyright (C) 2007,2008 Øyvind Harboe *
+* Copyright (C) 2007-2009 Øyvind Harboe *
* oyvind.har...@zylin.com *
* *
* This program is free software; you can redistribute it and/or modify *
@@ -663,37 +663,6 @@ void jtag_sleep(uint32_t us);
#define ERROR_JTAG_INIT_SOFT_FAIL(-110)
/**
- * jtag_add_dr_out() is a version of jtag_add_dr_scan() which
- * only scans data out. It operates on 32 bit integers instead
- * of 8 bit, which makes it a better impedance match with
- * the calling code which often operate on 32 bit integers.
- *
- * Current or end_state can not be TAP_RESET. end_state can be TAP_INVALID
- *
- * num_bits[i] is the number of bits to clock out from value[i] LSB first.
- *
- * If the device is in bypass, then that is an error condition in
- * the caller code that is not detected by this fn, whereas
- * jtag_add_dr_scan() does detect it. Similarly if the device is not in
- * bypass, data must be passed to it.
- *
- * If anything fails, then jtag_error will be set and jtag_execute() will
- * return an error. There is no way to determine if there was a failure
- * during this function call.
- *
- * This is an inline fn to speed up embedded hosts. Also note that
- * interface_jtag_add_dr_out() can be a *small* inline function for
- * embedded hosts.
- *
- * There is no jtag_add_dr_outin() version of this fn that also allows
- * clocking data back in. Patches gladly accepted!
- */
-void jtag_add_dr_out(struct jtag_tap* tap,
- int num_fields, const int* num_bits, const uint32_t* value,
- tap_state_t end_state);
-
-
-/**
* Set the current JTAG core execution error, unless one was set
* by a previous call previously. Driver or application code must
* use jtag_error_clear to reset jtag_error once this routine has been
@@ -725,4 +694,8 @@
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
+ +/* The minidriver contains inline versions of JTAG fn's */ +#include minidriver.h + This is bad, as you are creating a new layering violation that will need to be removed. You should move this #include to somewhere other than what should be our public API, probably inside the C files that need it. Presently, this change exposes _more_ internals, not less. I'm not quite sure how you define a layer violation here. How can I tell that there is a violation happening? A minidriver must be able to inline jtag_add_dr_out() into inner loops, this is what broke in the refactoring. All the rest of the refactoring is great stuff! jtag_add_dr_out() is also at the same time an API, so what was done in minidrivers was a carefully controlled layer violation that took advantage of the tight coupling between the CPU and the hardware = two pokes for some jtag_add_dr_out() cases. The jtag_add_dr_out() API was engineered with this specifically in mind. Do you have any thoughts on how to get back the functionality of letting the minidriver provide an inline version of jtag_add_dr_out()? Inlining is important because, not because of the fn call overhead, but because the arguments to jtag_add_dr_out() are often constants, so there is *significant* constant propagation going on that will reduce jtag_add_dr_out() to two pokes in many cases. Important on low latency low processing power(embedded) environments. -- Øyvind Harboe US toll free 1-866-980-3434 / International +47 51 63 25 00 http://www.zylin.com/zy1000.html ARM7 ARM9 ARM11 XScale Cortex JTAG debugger and flash programmer ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
I just noticed that there may be other problems with this; I question
the need for both -I and #include dirname/ Seems like one or the
other is enough, so there appears to be a new redundancy here.
On Wed, 2009-12-02 at 13:39 -0800, Zach Welch wrote:
Do not apply this until you fix the problem noted below.
Thanks,
Zach
On Wed, 2009-12-02 at 14:38 +0100, Øyvind Harboe wrote:
Fix 'regression' where jtag_add_dr_out() is no longer inlined
after refactoring.
The minidrivers strip away all the overhead and allow inner loops
to communicate directly with a hardware FIFO.
Signed-off-by: Øyvind Harboe oyvind.har...@zylin.com
---
src/jtag/Makefile.am |2 -
src/jtag/core.c | 14 -
src/jtag/driver.c| 10 +++
src/jtag/jtag.h | 37 +++--
src/jtag/minidriver.h| 67
++---
src/jtag/zy1000/zy1000.c |3 +-
6 files changed, 73 insertions(+), 60 deletions(-)
diff --git a/src/jtag/Makefile.am b/src/jtag/Makefile.am
index 5254a2b..c762300 100644
--- a/src/jtag/Makefile.am
+++ b/src/jtag/Makefile.am
@@ -11,11 +11,9 @@ if MINIDRIVER
if ZY1000
DRIVERFILES += zy1000/zy1000.c
-AM_CPPFLAGS += -I$(srcdir)/zy1000
endif
if MINIDRIVER_DUMMY
DRIVERFILES += minidummy/minidummy.c commands.c
-AM_CPPFLAGS += -I$(srcdir)/minidummy
endif
else
diff --git a/src/jtag/core.c b/src/jtag/core.c
index 9230cc2..cf14394 100644
--- a/src/jtag/core.c
+++ b/src/jtag/core.c
@@ -502,20 +502,6 @@ void jtag_add_plain_dr_scan(int in_num_fields, const
struct scan_field *in_field
jtag_set_error(retval);
}
-void jtag_add_dr_out(struct jtag_tap* tap,
- int num_fields, const int* num_bits, const uint32_t* value,
- tap_state_t end_state)
-{
- assert(end_state != TAP_RESET);
- assert(end_state != TAP_INVALID);
-
- cmd_queue_cur_state = end_state;
-
- interface_jtag_add_dr_out(tap,
- num_fields, num_bits, value,
- end_state);
-}
-
void jtag_add_tlr(void)
{
jtag_prelude(TAP_RESET);
diff --git a/src/jtag/driver.c b/src/jtag/driver.c
index cadd88e..fe56369 100644
--- a/src/jtag/driver.c
+++ b/src/jtag/driver.c
@@ -525,3 +525,13 @@ void interface_jtag_add_callback(jtag_callback1_t
callback, jtag_callback_data_t
jtag_add_callback4(jtag_convert_to_callback4, data0,
(jtag_callback_data_t)callback, 0, 0);
}
+void interface_jtag_alloc_in_value32(struct scan_field *field)
+{
+ field-in_value = (uint8_t *)cmd_queue_alloc(4);
+}
+
+void interface_jtag_add_scan_check_alloc(struct scan_field *field)
+{
+ unsigned num_bytes = DIV_ROUND_UP(field-num_bits, 8);
+ field-in_value = (uint8_t *)cmd_queue_alloc(num_bytes);
+}
diff --git a/src/jtag/jtag.h b/src/jtag/jtag.h
index ee96775..4b0e8b2 100644
--- a/src/jtag/jtag.h
+++ b/src/jtag/jtag.h
@@ -2,7 +2,7 @@
* Copyright (C) 2005 by Dominic Rath*
* dominic.r...@gmx.de *
* *
-* Copyright (C) 2007,2008 Øyvind Harboe *
+* Copyright (C) 2007-2009 Øyvind Harboe *
* oyvind.har...@zylin.com *
* *
* This program is free software; you can redistribute it and/or modify *
@@ -663,37 +663,6 @@ void jtag_sleep(uint32_t us);
#define ERROR_JTAG_INIT_SOFT_FAIL(-110)
/**
- * jtag_add_dr_out() is a version of jtag_add_dr_scan() which
- * only scans data out. It operates on 32 bit integers instead
- * of 8 bit, which makes it a better impedance match with
- * the calling code which often operate on 32 bit integers.
- *
- * Current or end_state can not be TAP_RESET. end_state can be TAP_INVALID
- *
- * num_bits[i] is the number of bits to clock out from value[i] LSB first.
- *
- * If the device is in bypass, then that is an error condition in
- * the caller code that is not detected by this fn, whereas
- * jtag_add_dr_scan() does detect it. Similarly if the device is not in
- * bypass, data must be passed to it.
- *
- * If anything fails, then jtag_error will be set and jtag_execute() will
- * return an error. There is no way to determine if there was a failure
- * during this function call.
- *
- * This is an inline fn to speed up embedded hosts. Also note that
- * interface_jtag_add_dr_out() can be a *small* inline function for
- * embedded hosts.
- *
- * There is no jtag_add_dr_outin() version of this fn that also allows
- * clocking data back in. Patches gladly accepted!
- */
-void jtag_add_dr_out(struct jtag_tap*
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Wed, Dec 2, 2009 at 10:55 PM, Zach Welch z...@superlucidity.net wrote: I just noticed that there may be other problems with this; I question the need for both -I and #include dirname/ Seems like one or the other is enough, so there appears to be a new redundancy here. I removed the -I and used only dirname/. I find the addition of numerous -I's confusing and error prone + makes it harder to read check build logs. -- Øyvind Harboe US toll free 1-866-980-3434 / International +47 51 63 25 00 http://www.zylin.com/zy1000.html ARM7 ARM9 ARM11 XScale Cortex JTAG debugger and flash programmer ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Wed, 2009-12-02 at 23:00 +0100, Øyvind Harboe wrote: On Wed, Dec 2, 2009 at 10:55 PM, Zach Welch z...@superlucidity.net wrote: I just noticed that there may be other problems with this; I question the need for both -I and #include dirname/ Seems like one or the other is enough, so there appears to be a new redundancy here. I removed the -I and used only dirname/. Thanks! I find the addition of numerous -I's confusing and error prone + makes it harder to read check build logs. Try my 'hdr' or 'tests' branches on my mirror. They remove almost all of the internal -I's from the tree!!! :) --Z ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Wed, 2009-12-02 at 22:54 +0100, Øyvind Harboe wrote: + +/* The minidriver contains inline versions of JTAG fn's */ +#include minidriver.h + This is bad, as you are creating a new layering violation that will need to be removed. You should move this #include to somewhere other than what should be our public API, probably inside the C files that need it. Presently, this change exposes _more_ internals, not less. Let me emphasize my point above: remove the #include from jtag.h. Do not expose it outside the JTAG layer. If it is a public API too, then you must rename the inline version and call it from a new wrapper function in jtag/core.c. You are exposing all of the minidriver to the entire system, and that's unacceptable. If this hurts performance, then please show us the numbers here first. Even then, this needs to be done in a way that maintains encapsulation. I'm not quite sure how you define a layer violation here. How can I tell that there is a violation happening? jtag.h public APIs for use by external applications minidriver.h internal APIs. These drivers will _always_ be built-in to the tree (for performance reasons). Thus, their interfaces should __never__ be exposed to users through the public JTAG API. minidummy/jtag_interface.h implementation APIs are even more internal, yet your patch will make these available in the public API!! No one should be adding new internals into public headers. If this means we need to create a slew of module_imp.h files to share bits of the internal implementation, then so be it. It's worth it, if it means that we can stop expose our internals. Nevertheless, you have to stop and really consider the consequences of stick stuff in headers! Who will see it? Should they?! ;) Create a new function to wrap the inline call if needed, but stop sticking stuff in the wrong place!! I can make our libraries usable for 0.4.0, though the APIs will not be stable for some time yet. Still, it opens the door to writing external applications; that option is closed today, and we are hurting the code by delaying. We'll never find out the problems until we start writing applications that use the code -- such as a test suite. Given these hard constraints, I recommend taking another stab at the patch and finding a way around this. To be honest, this might reflect fundamental design flaws with the minidriver, such that it can only be fixed after some rather drastic re-writing of the code. If the code relies this heavily on these optimizations, it's broken by design. Cheers, --Z ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Wed, 2009-12-02 at 14:32 -0800, Zach Welch wrote: On Wed, 2009-12-02 at 22:54 +0100, Øyvind Harboe wrote: + +/* The minidriver contains inline versions of JTAG fn's */ +#include minidriver.h + This is bad, as you are creating a new layering violation that will need to be removed. You should move this #include to somewhere other than what should be our public API, probably inside the C files that need it. Presently, this change exposes _more_ internals, not less. Let me emphasize my point above: remove the #include from jtag.h. Do not expose it outside the JTAG layer. If it is a public API too, then you must rename the inline version and call it from a new wrapper function in jtag/core.c. You are exposing all of the minidriver to the entire system, and that's unacceptable. [snip] Given these hard constraints, I recommend taking another stab at the patch and finding a way around this. To be honest, this might reflect fundamental design flaws with the minidriver, such that it can only be fixed after some rather drastic re-writing of the code. If the code relies this heavily on these optimizations, it's broken by design. Here's one possible solution: we could #include minidriver.h -- as long as it is an empty file when installed. This means two versions of minidriver.h, one that includes the proper jtag_minidriver.h and one that is empty. In fact, I think you can rename the later to the former name, then use the -I trick. Then add a mininone/minidriver.h that is the empty file, which is included when the others aren't. Grok? -- Z ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
I was thinking about splitting this patch into at least a few: The first would be to get rid of some -I's. The second would clarify that the classic JTAG driver is just another minidriver. The third would then allow a minidriver to grab the JTAG stack from jtag_xxx() and down to the metal. The performance gain potential here is, over time with a few more fixes like it, ca. 10-25%. I currently get 550kBytes/s @ 16MHz w/arm7/9 for gdb load, so say up to 600-700kBytes/s somewhere eventually due to this. Performance is a long term fun project for me :-) I actually saw that jtag_add_dr_out() was no longer inlined because it is creeping up on the gprof list for gdb load... -- Øyvind Harboe US toll free 1-866-980-3434 / International +47 51 63 25 00 http://www.zylin.com/zy1000.html ARM7 ARM9 ARM11 XScale Cortex JTAG debugger and flash programmer ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
On Wed, 2009-12-02 at 14:44 -0800, Zach Welch wrote: On Wed, 2009-12-02 at 14:32 -0800, Zach Welch wrote: On Wed, 2009-12-02 at 22:54 +0100, Øyvind Harboe wrote: + +/* The minidriver contains inline versions of JTAG fn's */ +#include minidriver.h + This is bad, as you are creating a new layering violation that will need to be removed. You should move this #include to somewhere other than what should be our public API, probably inside the C files that need it. Presently, this change exposes _more_ internals, not less. Let me emphasize my point above: remove the #include from jtag.h. Do not expose it outside the JTAG layer. If it is a public API too, then you must rename the inline version and call it from a new wrapper function in jtag/core.c. You are exposing all of the minidriver to the entire system, and that's unacceptable. [snip] Given these hard constraints, I recommend taking another stab at the patch and finding a way around this. To be honest, this might reflect fundamental design flaws with the minidriver, such that it can only be fixed after some rather drastic re-writing of the code. If the code relies this heavily on these optimizations, it's broken by design. Here's one possible solution: we could #include minidriver.h -- as long as it is an empty file when installed. This means two versions of minidriver.h, one that includes the proper jtag_minidriver.h and one that is empty. In fact, I think you can rename the later to the former name, then use the -I trick. Then add a mininone/minidriver.h that is the empty file, which is included when the others aren't. Grok? I am probably wrong about some details (e.g. renaming), but the goal of this is to eliminate the #if logic from the header files too. That is not acceptable for installed libraries either, so we should not be adding more of that to headers The tricks above are similar to those that I used to clean openocd.c of the same logic, so those efforts might help provide you with the required inspiration to solve this. --Z ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
I am probably wrong about some details (e.g. renaming), but the goal of this is to eliminate the #if logic from the header files too. That is not acceptable for installed libraries either, so we should not be adding more of that to headers The tricks above are similar to those that I used to clean openocd.c of the same logic, so those efforts might help provide you with the required inspiration to solve this. The minidrivers and inlining of jtag_add_dr_out() is pulling off something that's inheritely incestuous. The active minidriver must be included conditionally by some trick or another into all users of the jtag_add_xxx() API. -- Øyvind Harboe US toll free 1-866-980-3434 / International +47 51 63 25 00 http://www.zylin.com/zy1000.html ARM7 ARM9 ARM11 XScale Cortex JTAG debugger and flash programmer ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
Re: [Openocd-development] [PATCH] minidriver: allow jtag_add_dr_out() to be inlined into inner loops
Okay. The solution that I have tried to propose will get this back, without exposing anything during a normal non-minidriver installation. The point I'm trying to make is that there is no normal non-minidriver. The normal/classic stack is *also* a minidriver. The *first* step would be to split out the normal non-minidriver as a minidriver just like the rest. That should make things more clear. Essentially the minidriver supports *everything* from jtag_add_xxx() API and down to the metal. This goes for the classic case as well. I won't be pushing this issue with any haste, especially as there is lots of stuff in flight. -- Øyvind Harboe US toll free 1-866-980-3434 / International +47 51 63 25 00 http://www.zylin.com/zy1000.html ARM7 ARM9 ARM11 XScale Cortex JTAG debugger and flash programmer ___ Openocd-development mailing list Openocd-development@lists.berlios.de https://lists.berlios.de/mailman/listinfo/openocd-development
