Re: [ECOS] Use of exception and RTTI in application C+ code

[Jonathan Larmour <jifl at eCosCentric dot com>]

W. Jensen wrote:
> Jonathan,
>
> Can you send me this patch when you get a chance? I'm curious about what is
> involved.  I assume that at the very least it needs an exception stack for
> each thread.  Perhaps I can port it to the CF platform if it is not too
> complicated.  I am a bit concerned that it would be a gcc-version-dependant
> patch though.  If that is the case, maybe it is not worth it.
It is gcc 3.x only. In fact at the time it was done, it was early 3.x so I 
can't guarantee changes aren't required for more recent gcc.

But there is a possibility it may be easy for you to get something going 
for yourself in some cases in fact. For your information, and _only_ if 
you need something for yourself here's the type of change required. It's 
quite simple in this case. This .ldi file gives an example of what you may 
need to add to your own memory layout:

Index: include/pkgconf/mlt_arm_ebsa285_ram.ldi
===================================================================
--- include/pkgconf/mlt_arm_ebsa285_ram.ldi
+++ include/pkgconf/mlt_arm_ebsa285_ram.ldi
@@ -12,17 +12,20 @@ MEMORY
 SECTIONS
 {
     SECTIONS_BEGIN
     SECTION_fixed_vectors (ram, 0x20, LMA_EQ_VMA)
     SECTION_rom_vectors (ram, 0x20000, LMA_EQ_VMA)
+    SECTION_RELOCS (ram, ALIGN (0x4), LMA_EQ_VMA)
     SECTION_text (ram, ALIGN (0x4), LMA_EQ_VMA)
     SECTION_fini (ram, ALIGN (0x4), LMA_EQ_VMA)
     SECTION_rodata (ram, ALIGN (0x4), LMA_EQ_VMA)
     SECTION_rodata1 (ram, ALIGN (0x4), LMA_EQ_VMA)
     SECTION_fixup (ram, ALIGN (0x4), LMA_EQ_VMA)
     SECTION_gcc_except_table (ram, ALIGN (0x4), LMA_EQ_VMA)
+    SECTION_eh_frame (ram, ALIGN (0x4), LMA_EQ_VMA)
     SECTION_data (ram, ALIGN (0x4), LMA_EQ_VMA)
+    SECTION_got (ram, ALIGN (0x4), LMA_EQ_VMA)
     SECTION_bss (ram, ALIGN (0x4), LMA_EQ_VMA)
     CYG_LABEL_DEFN(__heap1) = ALIGN (0x8);
     CYG_LABEL_DEFN(__pci_window) = 0xf00000; . = 
CYG_LABEL_DEFN(__pci_window) + 0x100000;
     SECTIONS_END
 }

Index: pkgconf/rules.mak
===================================================================
--- pkgconf/rules.mak
+++ pkgconf/rules.mak
@@ -120,13 +120,13 @@ ifeq ($(HOST),CYGWIN)
 	@mkdir -p `cygpath -w "$(dir $@)" | sed "s/\\\\\/\\//g"`
 else
 	@mkdir -p $(dir $@)
 endif	
 ifneq ($(IGNORE_LINK_ERRORS),)
-	-$(CC) $(LDFLAGS) -L$(PREFIX)/lib -Ttarget.ld -o $@ $(<:.d=.o)
+	-$(CC) -L$(PREFIX)/lib -Ttarget.ld -o $@ $(<:.d=.o) $(LDFLAGS)
 else
-	$(CC) $(LDFLAGS) -L$(PREFIX)/lib -Ttarget.ld -o $@ $(<:.d=.o)
+	$(CC) -L$(PREFIX)/lib -Ttarget.ld -o $@ $(<:.d=.o) $(LDFLAGS)
 endif

 # rule to generate all tests and create a dependency file "tests.deps" by
 # concatenating the individual dependency rule files (file extension = ".d")
 # generated during compilation


Some of the other required changes are already included in arm.ld, but not 
for other architectures. The changes are non-trivial, and I don't really 
have the time to detail them right at the moment.

There is much more to it than this already that has been written but not 
checked in. On that basis I can't really accept a patch to do it - it 
requires more work than this relatively trivial excerpt I show above. For 
example the resulting exceptions will _not_ be thread safe. That work is 
almost complete, but also not complete as it requires gcc changes that 
have not yet been made.

So use it for yourself, if it works. But it may not, and if not then 
"caveat emptor", sorry.

Jifl
--
eCosCentric       http://www.eCosCentric.com/       <info@eCosCentric.com>
--[ "You can complain because roses have thorns, or you ]--
--[  can rejoice because thorns have roses." -Lincoln   ]-- Opinions==mine


--
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss