Commit Date  
[r142] by ma-work

Clean up after build, unless $SUPPRESS_CLEAN is nonempty.

2010-09-13 15:12:16 Tree
[r141] by ma-work

Build GDB after Insight, because Insight also installs gdb[tui].

2010-09-03 14:59:33 Tree
[r140] by ma-work

Remove the fragile curses.h check.

2010-09-03 14:58:48 Tree
[r139] by pabigot

Give up on being clever

2010-07-01 23:04:39 Tree
[r138] by pabigot

Try two: previous doesn't update sudo's path, which was the whole point.

2010-05-30 21:21:37 Tree
[r137] by pabigot

Avoid problems when INSTALL_LAUNCHER is not sudo

2010-05-30 19:47:23 Tree
[r136] by pabigot

Add file missed in commit 133

2010-05-20 19:17:08 Tree
[r135] by pabigot

Add file missed in commit 132

2010-05-20 19:16:19 Tree
[r134] by pabigot

Add new libc releases; correct host project. Avoid problems if target
directory not in sudo path.

2010-05-20 19:05:54 Tree
[r133] by pabigot

Revise generic categorization to match binutils and support new architectures.

It appears that the generic MSP430 configurations for gcc (msp1 and msp2)
are derived from original binutils generic configurations. Current binutils
(2.19.1 and 2.20) have six generic configurations, which map to the
following chip descriptions:

msp430x110: {"msp1", MSP430_ISA_11, 0}, /* anything without hardware multiply */
msp430x336: {"msp2", MSP430_ISA_14, 1}, /* older chips with hardware multiply */
msp430xG4616: {"msp3", MSP430_ISA_46, 1},
msp430x4783: {"msp4", MSP430_ISA_47, 1}, MSP430 architecture
msp430x47166: {"msp5", MSP430_ISA_471, 1}, MSP430X architecture
msp430x5418: {"msp6", MSP430_ISA_54, 1}, MSP430XV2 architecture

Next to each in this table is a specific chip that is taken as an example
for that family. The first two are the ones used previously in GCC; I don't
know why msp430x336 was used as the example for msp2 instead of something
like msp430x147. The remainder are simply the first listed chip of the
corresponding category.

Since binutils considers these to be the suite of generic architectures, I
decided to update the gcc patches to be in sync. This required changes in
several places:

msp430.c:msp430_mcu_types was extended with the additional generic types

msp430.h:CPP_SPEC, LINK_SPEC, and CRT_BINUTILS_SPECS were revised to include
the generics

I also removed the preprocessor definitions that had been added to certain
lines in CPP_SPEC, and instead placed those within the CPP_mlib_SPEC define
for the corresponding generic. This revealed that there are two macros used
to indicate the presence of a hardware multiply capability. Where I ended
up removing chip-specific defines, I verified that the newly assigned
generic chip matched those defines; I double-checked this against the TI
headers (e.g., ISA_47 chips are not MSP430X, but ISA_471 are).

Most of this last was aided by a Python script that I used to generate the
corresponding blocks of code, to make sure specific chips were not dropped
(e.g., some had been missing from LINK_SPEC).

2010-05-20 19:04:47 Tree
[r132] by pabigot

Update to binutils 2.20.1.

The patch for 2.20 applied fairly cleanly, with only a little fuzz; there
were no changes necessary for this new binutils version.

2010-05-20 19:02:14 Tree
[r131] by ma-work

Validate environment for typical bootstrapping pitfalls (libncurses*-dev, texinfo.)

2010-04-20 14:15:38 Tree
[r130] by ma-work

Fix X Y --fetch-only, and omitted OVERRIDE_URL.

2010-04-15 14:40:56 Tree
[r129] by ma-work

Minor formatting improvements of prompts.

2010-02-22 18:01:42 Tree
[r128] by ma-work

Add .gitignore file for the benefit of git-svn users.

2010-02-22 15:58:13 Tree
[r127] by ma-work

Offer option to strip output binaries.

2010-02-22 15:58:05 Tree
[r126] by ma-work

Support sudo. Rename $1 to $DEST.

2010-02-22 15:57:53 Tree
[r125] by ma-work

Update and revise a bit.

2010-02-22 15:57:45 Tree
[r124] by ma-work

Get gcc and gmp files in one wget call (for connection reuse).

2010-02-22 15:57:37 Tree
[r123] by ma-work

Fix tar when running @COMMANDS directly...

- the cd command is ineffective if in a system() call separate from tar
- also pipe into bzip2 -c, in case tar doesn't understand the j option.
- omit the final non-effective cd when executing script directly.

In the long run, we may want to pipe the @COMMANDS into a shell, it's easier
and reduces inconsistencies like these.

2010-02-22 15:08:53 Tree
[r122] by ma-work

Fix exit logic when running @COMMANDS directly, and...

...also use SystemCheck to report errors in more detail.

2010-02-22 15:08:46 Tree
[r121] by ma-work

Check for dialog(1) errors properly.

2010-02-22 15:08:38 Tree
[r120] by ma-work

Band-aid: re-run libc build without parallel make after failure.

2010-02-22 14:12:03 Tree
[r119] by ma-work

Get initial cd(1) right no matter where we start...

and use the dirname of $0, the currently running script.

2010-02-22 14:11:41 Tree
[r118] by ma-work

Avoid re-fetching libc, use wget -c. Suggested by Peter Bigot.

2010-02-22 14:11:16 Tree
Older >

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:

No, thanks