From: Allen Martin <hidden> Date: 2012-08-29 19:13:10
The LIST_arm rule included the Atmel boards twice (by virtue of
including both LIST_at91 and LIST_ARM9) and was missing all the
arm720t, arm946es, and arm1176 boards. Change this list to use
boards_by_arch() which is less error prone. After this change
"./MAKEALL arm" and "./MAKEALL -a arm" build the same boards.
Also add the missing arm946es and arm1176 processors to the ARM9
and ARM11 lists respectively and add an ARM7 list for the arm720t
processor.
Signed-off-by: Allen Martin <redacted>
---
MAKEALL | 22 +++++++++++-----------
1 file changed, 11 insertions(+), 11 deletions(-)
The LIST_arm rule included the Atmel boards twice (by virtue of
including both LIST_at91 and LIST_ARM9) and was missing all the
arm720t, arm946es, and arm1176 boards. Change this list to use
boards_by_arch() which is less error prone. After this change
"./MAKEALL arm" and "./MAKEALL -a arm" build the same boards.
Also add the missing arm946es and arm1176 processors to the ARM9
and ARM11 lists respectively and add an ARM7 list for the arm720t
processor.
Good find! Would you mind looking quickly over the other arches to see
if they have any similar but now obvious problems (where LIST_${ARCH}
doesn't use boards_by_arch)) ? Thanks!
--
Tom
From: Allen Martin <hidden> Date: 2012-08-29 20:13:40
On Wed, Aug 29, 2012 at 01:03:40PM -0700, Tom Rini wrote:
On 08/29/2012 12:13 PM, Allen Martin wrote:
quoted
The LIST_arm rule included the Atmel boards twice (by virtue of
including both LIST_at91 and LIST_ARM9) and was missing all the
arm720t, arm946es, and arm1176 boards. Change this list to use
boards_by_arch() which is less error prone. After this change
"./MAKEALL arm" and "./MAKEALL -a arm" build the same boards.
Also add the missing arm946es and arm1176 processors to the ARM9
and ARM11 lists respectively and add an ARM7 list for the arm720t
processor.
Good find! Would you mind looking quickly over the other arches to see
if they have any similar but now obvious problems (where LIST_${ARCH}
doesn't use boards_by_arch)) ? Thanks!
--
Tom
By experiment it look like m68k and mips have problems too, I'll investigate:
$ for i in $(awk '($1 !~ /^#/) {print $2}' boards.cfg | sort | uniq) ; do
From: Allen Martin <hidden> Date: 2012-08-29 20:54:07
On Wed, Aug 29, 2012 at 01:13:39PM -0700, Allen Martin wrote:
On Wed, Aug 29, 2012 at 01:03:40PM -0700, Tom Rini wrote:
quoted
On 08/29/2012 12:13 PM, Allen Martin wrote:
quoted
The LIST_arm rule included the Atmel boards twice (by virtue of
including both LIST_at91 and LIST_ARM9) and was missing all the
arm720t, arm946es, and arm1176 boards. Change this list to use
boards_by_arch() which is less error prone. After this change
"./MAKEALL arm" and "./MAKEALL -a arm" build the same boards.
Also add the missing arm946es and arm1176 processors to the ARM9
and ARM11 lists respectively and add an ARM7 list for the arm720t
processor.
Good find! Would you mind looking quickly over the other arches to see
if they have any similar but now obvious problems (where LIST_${ARCH}
doesn't use boards_by_arch)) ? Thanks!
--
Tom
By experiment it look like m68k and mips have problems too, I'll investigate:
It looks like mips splits out the little endian boards into a virtual
"mips_el" architecture that doesn't actually exist in boardfs.cfg, so
maybe I won't touch that. I founds some other missing/duplicate
boards in mips and m68k though so I'll add those to the patch.
-Allen
--
nvpublic