Support for ATEVK1100 evaluation board
Subsystems (4)
v2 cover Support for ATEVK1100 evaluation board
This is a patch series which adds support for the ATEVK1100 evaluation board[1], and the AT32UC3A0xxx[2] microcontrollers used on that board. The patch series is based on avr32/next.
Some of these patches has already been submitted and most of the received feedback has been addressed. We would like to thank you for all the received feedback, and say thay we are sorry for the long delay.
This microcontroller is similar to the AT32AP700x, which is already supported by U-Boot, and this code is therefore based on that.
Patch 1 makes a change which works around a bug in the compiler. Patch 2-7 makes changes to other avr32-code in preparation for support of the new microcontroller.
Patch 8 adds support for the AT32UC3A0xxx microcontrollers.
Patch 9 adds support for the ATEVK1100 evaluation board.
The current revisions of the microcontroller have a bug in the SDRAM-controller, which makes it unreliable to run code from SDRAM. We therefore currently don't use SDRAM, but rather SRAM.
Changes since last time:
- Merged the two patch-series.
- Config option for searching for phy-address.
- Config option for limiting macb speed to 10 Mbps.
- Changed serial port speed to 115200.
- Some cleanups.
What works:
Loading U-Boot, serial console, loading image over dhcp/tftp, booting image.
[1] http://www.atmel.com/dyn/Products/tools_card.asp?tool_id=4114
[2] http://www.atmel.com/products/avr32/uc3/uc3_2.asp?family_id=682
PATCH v2 1/9 Fix IP alignement problem
From: Olav Morken <[email protected]>
This patch removes volatile from:
volatile IP_t *ip = (IP_t *)xip;
Due to a bug, avr32-gcc will assume that ip is aligned on a word boundary when using volatile, which causes an exception since xip isn't aligned on a word boundary.
--- net/net.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
net/net.c +1 -1
@@ -1685,7 +1685,7 @@ NetSetEther(volatile uchar * xet, uchar * addr, uint prot)void NetSetIP (volatile uchar * xip,IPaddr_t dest,int dport,int sport,int len) { -volatile IP_t *ip = (IP_t *)xip; +IP_t *ip = (IP_t *)xip;/* * If the data is an odd number of bytes, zero the
Signature · 1 line
1.6.0.4
PATCH v2 2/9 AVR32: Make cacheflush cpu-dependent
From: Olav Morken <[email protected]>
The AT32UC3A series of processors doesn't contain any cache, and issuing cache control instructions on those will cause an exception. This commit makes cacheflush.h arch-dependent in preparation for the AT32UC3A-support.
---
board/atmel/atstk1000/flash.c | 2 +-
board/earthlcd/favr-32-ezkit/flash.c | 2 +-
cpu/at32ap/cache.c | 2 +-
.../asm-avr32/{ => arch-at32ap700x}/cacheflush.h | 0
include/asm-avr32/dma-mapping.h | 2 +-
lib_avr32/board.c | 2 +-
6 files changed, 5 insertions(+), 5 deletions(-)
rename include/asm-avr32/{ => arch-at32ap700x}/cacheflush.h (100%)board/atmel/atstk1000/flash.c +1 -1
@@ -22,7 +22,7 @@#include <common.h> #ifdef CONFIG_ATSTK1000_EXT_FLASH -#include <asm/cacheflush.h> +#include <asm/arch/cacheflush.h> #include <asm/io.h> #include <asm/sections.h>
board/earthlcd/favr-32-ezkit/flash.c +1 -1
@@ -20,7 +20,7 @@#include <common.h> #ifdef CONFIG_FAVR32_EZKIT_EXT_FLASH -#include <asm/cacheflush.h> +#include <asm/arch/cacheflush.h> #include <asm/io.h> #include <asm/sections.h>
cpu/at32ap/cache.c +1 -1
@@ -22,7 +22,7 @@#include <common.h> -#include <asm/cacheflush.h> +#include <asm/arch/cacheflush.h> void dcache_clean_range (volatile void *start,size_t size) {
include/asm-avr32/arch-at32ap700x/cacheflush.h renamed from include/asm-avr32/cacheflush.h +0 -0
include/asm-avr32/dma-mapping.h +1 -1
@@ -23,7 +23,7 @@#define __ASM_AVR32_DMA_MAPPING_H#include <asm/io.h> -#include <asm/cacheflush.h> +#include <asm/arch/cacheflush.h> enum dma_data_direction {DMA_BIDIRECTIONAL =0 ,
lib_avr32/board.c +1 -1
@@ -85,7 +85,7 @@ void *sbrk(ptrdiff_t increment) }#ifdef CFG_DMA_ALLOC_LEN -#include <asm/cacheflush.h> +#include <asm/arch/cacheflush.h> #include <asm/io.h> static unsigned long dma_alloc_start;
Signature · 1 line
1.6.0.4
PATCH v2 3/9 AVR32: Move addrspace.h to arch-directory, and move some functions from io.h to addrspace.h
From: Olav Morken <[email protected]>
The AVR32A architecture (which AT32UC3A-series is based on) has a different memory layout than the AVR32B-architecture. This patch moves addrspace.h to an arch-dependent directory in preparation for AT32UC3A-support. It also moves some address-space manipulation functions from io.h to addrspace.h.
--- include/asm-avr32/addrspace.h | 46 -------------- include/asm-avr32/arch-at32ap700x/addrspace.h | 84 +++++++++++++++++++++++++ include/asm-avr32/io.h | 39 +----------- lib_avr32/bootm.c | 2 +- 4 files changed, 87 insertions(+), 84 deletions(-) delete mode 100644 include/asm-avr32/addrspace.h create mode 100644 include/asm-avr32/arch-at32ap700x/addrspace.h
include/asm-avr32/addrspace.h deleted +0 -46
@@ -1,46 +0,0 @@ -/* -* Copyright (C) 2006 Atmel Corporation -* -* See file CREDITS for list of people who contributed to this -* project. -* -* This program is free software; you can redistribute it and/or -* modify it under the terms of the GNU General Public License as -* published by the Free Software Foundation; either version 2 of -* the License, or (at your option) any later version. -* -* This program is distributed in the hope that it will be useful, -* but WITHOUT ANY WARRANTY; without even the implied warranty of -* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the -* GNU General Public License for more details. -* -* You should have received a copy of the GNU General Public License -* along with this program; if not, write to the Free Software -* Foundation, Inc., 59 Temple Place, Suite 330, Boston, -* MA 02111-1307 USA -*/ -#ifndef __ASM_AVR32_ADDRSPACE_H -#define __ASM_AVR32_ADDRSPACE_H - -/* Memory segments when segmentation is enabled */ -#define P0SEG 0x00000000 -#define P1SEG 0x80000000 -#define P2SEG 0xa0000000 -#define P3SEG 0xc0000000 -#define P4SEG 0xe0000000 - -/* Returns the privileged segment base of a given address */ -#define PXSEG (a) (((unsigned long)(a)) & 0xe0000000) - -/* Returns the physical address of a PnSEG (n=1,2) address */ -#define PHYSADDR (a) (((unsigned long)(a)) & 0x1fffffff) - -/* -* Map an address to a certain privileged segment -*/ -#define P1SEGADDR (a) ((__typeof__(a))(((unsigned long)(a) & 0x1fffffff) | P1SEG)) -#define P2SEGADDR (a) ((__typeof__(a))(((unsigned long)(a) & 0x1fffffff) | P2SEG)) -#define P3SEGADDR (a) ((__typeof__(a))(((unsigned long)(a) & 0x1fffffff) | P3SEG)) -#define P4SEGADDR (a) ((__typeof__(a))(((unsigned long)(a) & 0x1fffffff) | P4SEG)) - -#endif /* __ASM_AVR32_ADDRSPACE_H */
include/asm-avr32/arch-at32ap700x/addrspace.h new file +84 -0
84 lines not on this page · raw message
include/asm-avr32/io.h +2 -37
@@ -73,21 +73,8 @@ extern void __readwrite_bug(const char *fn);#define inw (p) ({ unsigned int __v = __le16_to_cpu(__raw_readw(p)); __v; })#define inl (p) ({ unsigned int __v = __le32_to_cpu(__raw_readl(p)); __v; }) -#include <asm/addrspace.h> - -/* virt_to_phys will only work when address is in P1 or P2 */ -static __inline__unsigned long virt_to_phys (volatile void *address) -{ -return PHYSADDR (address); -} - -static __inline__void *phys_to_virt (unsigned long address) -{ -return (void *)P1SEGADDR (address); -} - -#define cached (addr) ((void *)P1SEGADDR(addr)) -#define uncached (addr) ((void *)P2SEGADDR(addr)) +#include <asm/arch/addrspace.h> +/* Provides virt_to_phys, phys_to_virt, cached, uncached, map_physmem */ #endif /* __KERNEL__ */
@@ -96,28 +83,6 @@ static inline void sync(void) }/* -* Given a physical address and a length, return a virtual address -* that can be used to access the memory range with the caching -* properties specified by "flags". -* -* This implementation works for memory below 512MiB (flash, etc.) as -* well as above 3.5GiB (internal peripherals.) -*/ -#define MAP_NOCACHE (0) -#define MAP_WRCOMBINE (1 << 7) -#define MAP_WRBACK (MAP_WRCOMBINE | (1 << 9)) -#define MAP_WRTHROUGH (MAP_WRBACK | (1 << 0)) - -static inline void * -map_physmem (phys_addr_t paddr,unsigned long len,unsigned long flags) -{ -if (flags ==MAP_WRBACK ) -return (void *)P1SEGADDR (paddr); -else -return (void *)P2SEGADDR (paddr); -} - -/* * Take down a mapping set up by map_physmem(). */ static inline void unmap_physmem (void *vaddr,unsigned long len)
lib_avr32/bootm.c +1 -1
@@ -24,7 +24,7 @@#include <image.h> #include <zlib.h> #include <asm/byteorder.h> -#include <asm/addrspace.h> +#include <asm/arch/addrspace.h> #include <asm/io.h> #include <asm/setup.h> #include <asm/arch/clk.h>
Signature · 1 line
1.6.0.4
PATCH v2 4/9 AVR32: Make GPIO implmentation cpu dependent
There are some differences in the implementation of GPIO in the at32uc chip compared to the ap700x series.
--- include/asm-avr32/arch-at32ap700x/gpio-impl.h | 86 +++++++++++++++++++++++++ include/asm-avr32/arch-common/portmux-gpio.h | 83 +----------------------- 2 files changed, 88 insertions(+), 81 deletions(-) create mode 100644 include/asm-avr32/arch-at32ap700x/gpio-impl.h
include/asm-avr32/arch-at32ap700x/gpio-impl.h new file +86 -0
@@ -0,0 +1,86 @@ +#ifndef __ASM_AVR32_ARCH_GPIO_IMPL_H__ +#define __ASM_AVR32_ARCH_GPIO_IMPL_H__ + +/* Register offsets */ +struct gpio_regs { +u32 GPER; +u32 GPERS; +u32 GPERC; +u32 GPERT; +u32 PMR0; +u32 PMR0S; +u32 PMR0C; +u32 PMR0T; +u32 PMR1; +u32 PMR1S; +u32 PMR1C; +u32 PMR1T; +u32 __reserved0[4 ]; +u32 ODER; +u32 ODERS; +u32 ODERC; +u32 ODERT; +u32 OVR; +u32 OVRS; +u32 OVRC; +u32 OVRT; +u32 PVR; +u32 __reserved_PVRS; +u32 __reserved_PVRC; +u32 __reserved_PVRT; +u32 PUER; +u32 PUERS; +u32 PUERC; +u32 PUERT; +u32 PDER; +u32 PDERS; +u32 PDERC; +u32 PDERT; +u32 IER; +u32 IERS; +u32 IERC; +u32 IERT; +u32 IMR0; +u32 IMR0S; +u32 IMR0C; +u32 IMR0T; +u32 IMR1; +u32 IMR1S; +u32 IMR1C; +u32 IMR1T; +u32 GFER; +u32 GFERS; +u32 GFERC; +u32 GFERT; +u32 IFR; +u32 __reserved_IFRS; +u32 IFRC; +u32 __reserved_IFRT; +u32 ODMER; +u32 ODMERS; +u32 ODMERC; +u32 ODMERT; +u32 __reserved1[4 ]; +u32 ODCR0; +u32 ODCR0S; +u32 ODCR0C; +u32 ODCR0T; +u32 ODCR1; +u32 ODCR1S; +u32 ODCR1C; +u32 ODCR1T; +u32 __reserved2[4 ]; +u32 OSRR0; +u32 OSRR0S; +u32 OSRR0C; +u32 OSRR0T; +u32 __reserved3[8 ]; +u32 STER; +u32 STERS; +u32 STERC; +u32 STERT; +u32 __reserved4[35 ]; +u32 VERSION; +}; + +#endif /* __ASM_AVR32_ARCH_GPIO_IMPL_H__ */
include/asm-avr32/arch-common/portmux-gpio.h +2 -81
89 lines not on this page · raw message
Signature · 1 line
1.6.0.4
PATCH v2 5/9 AVR32: macb - Disable 100mbps if clock is slow
From: Olav Morken <[email protected]>
For 100mbps operation, the ethernet controller requires a 25 MHz clock in MII mode, and a 50 MHz clock in RMII mode. If the clock is slower, disable 100 Mbps mode.
--- drivers/net/macb.c | 12 +++++++++++- 1 files changed, 11 insertions(+), 1 deletions(-)
drivers/net/macb.c +11 -1
@@ -297,7 +297,17 @@ static void macb_phy_reset(struct macb_device *macb)int i;u16 status, adv; - adv =ADVERTISE_CSMA |ADVERTISE_ALL ; + adv =ADVERTISE_CSMA |ADVERTISE_ALL ; + +#ifdef CONFIG_MACB_FORCE10M +printf ("%s: 100Mbps is not supported on this board - forcing 10Mbps.\n" , + netdev->name); + + adv &= ~ADVERTISE_100FULL ; + adv &= ~ADVERTISE_100HALF ; + adv &= ~ADVERTISE_100BASE4 ; +#endif +macb_mdio_write (macb,MII_ADVERTISE , adv);printf ("%s: Starting autonegotiation...\n" , netdev->name); macb_mdio_write(macb, MII_BMCR, (BMCR_ANENABLE
Signature · 1 line
1.6.0.4
PATCH v2 6/9 AVR32: macb - Search for PHY id
This patch adds support for searching through available PHY-addresses in the macb-driver. This is needed for the ATEVK1100 evaluation board, where the PHY-address will be initialized to either 1 or 7.
This patch adds a config option, CONFIG_MACB_SEARCH_PHY, which when enabled tells the driver to search for the PHY address.
--- drivers/net/macb.c | 31 +++++++++++++++++++++++++++++++ 1 files changed, 31 insertions(+), 0 deletions(-)
drivers/net/macb.c +31 -0
@@ -327,6 +327,30 @@ static void macb_phy_reset(struct macb_device *macb) netdev->name, status); } +#ifdef CONFIG_MACB_SEARCH_PHY +static int macb_phy_find (struct macb_device *macb) +{ +int i; +u16 phy_id; + +/* Search for PHY... */ +for (i =0 ; i <32 ; i++) { + macb->phy_addr = i; + phy_id =macb_mdio_read (macb,MII_PHYSID1 ); +if (phy_id !=0xffff ) { +printf ("%s: PHY present at %d\n" , macb->netdev.name, i); +return 1 ; + } + } + +/* PHY isn't up to snuff */ +printf ("%s: PHY not found" , macb->netdev.name); + +return 0 ; +} +#endif /* CONFIG_MACB_SEARCH_PHY */ + +static int macb_phy_init (struct macb_device *macb) {struct eth_device *netdev = &macb->netdev;
@@ -335,6 +359,13 @@ static int macb_phy_init(struct macb_device *macb)int media, speed, duplex;int i; +#ifdef CONFIG_MACB_SEARCH_PHY +/* Auto-detect phy_addr */ +if (!macb_phy_find (macb)) { +return 0 ; + } +#endif /* CONFIG_MACB_SEARCH_PHY */ +/* Check if the PHY is up to snuff... */ phy_id =macb_mdio_read (macb,MII_PHYSID1 );if (phy_id ==0xffff ) {
Signature · 1 line
1.6.0.4
PATCH v2 7/9 AVR32: Must add NOPs after disabling interrupts for AT32UC3A0512ES
From: Olav Morken <[email protected]>
The AT32UC3A0512ES chip has a bug when disabling interrupts. As a workaround, two NOPs can be inserted.
--- lib_avr32/interrupts.c | 7 +++++++ 1 files changed, 7 insertions(+), 0 deletions(-)
lib_avr32/interrupts.c +7 -0
@@ -35,5 +35,12 @@ int disable_interrupts(void) sr =sysreg_read (SR ); asmvolatile ("ssrf %0" : :"n" (SYSREG_GM_OFFSET )); +#ifdef CONFIG_AT32UC3A0xxx +/* Two NOPs are required after masking interrupts on the +* AT32UC3A0512ES. See errata 41.4.5.5. */ + asm("nop" ); + asm("nop" ); +#endif +return !SYSREG_BFEXT (GM , sr); }
Signature · 1 line
1.6.0.4
PATCH v2 8/9 AVR32: CPU support for AT32UC3A0xxx CPUs
This patch adds support for the AT32UC3A0xxx chips.
--- cpu/at32uc/Makefile | 57 +++++ cpu/at32uc/at32uc3a0xxx/Makefile | 43 ++++ cpu/at32uc/at32uc3a0xxx/clk.c | 88 +++++++ cpu/at32uc/at32uc3a0xxx/portmux.c | 154 ++++++++++++ cpu/at32uc/at32uc3a0xxx/sm.h | 247 +++++++++++++++++++ cpu/at32uc/cache.c | 34 +++ cpu/at32uc/config.mk | 22 ++ cpu/at32uc/cpu.c | 79 ++++++ cpu/at32uc/exception.c | 121 ++++++++++ cpu/at32uc/flashc.c | 144 +++++++++++ cpu/at32uc/flashc.h | 84 +++++++ cpu/at32uc/interrupts.c | 150 ++++++++++++ cpu/at32uc/portmux-gpio.c | 78 ++++++ cpu/at32uc/sdramc.c | 120 ++++++++++ cpu/at32uc/sdramc.h | 137 +++++++++++ cpu/at32uc/smc.c | 61 +++++ cpu/at32uc/smc.h | 105 ++++++++ cpu/at32uc/start.S | 252 ++++++++++++++++++++ include/asm-avr32/arch-at32uc3a0xxx/addrspace.h | 63 +++++ include/asm-avr32/arch-at32uc3a0xxx/cacheflush.h | 41 ++++ .../asm-avr32/arch-at32uc3a0xxx/chip-features.h | 30 +++ include/asm-avr32/arch-at32uc3a0xxx/clk.h | 85 +++++++ include/asm-avr32/arch-at32uc3a0xxx/gpio-impl.h | 62 +++++ include/asm-avr32/arch-at32uc3a0xxx/gpio.h | 44 ++++ include/asm-avr32/arch-at32uc3a0xxx/hmatrix.h | 47 ++++ include/asm-avr32/arch-at32uc3a0xxx/memory-map.h | 75 ++++++ include/asm-avr32/arch-at32uc3a0xxx/portmux.h | 80 ++++++ include/asm-avr32/sram.h | 34 +++ 28 files changed, 2537 insertions(+), 0 deletions(-) create mode 100644 cpu/at32uc/Makefile create mode 100644 cpu/at32uc/at32uc3a0xxx/Makefile create mode 100644 cpu/at32uc/at32uc3a0xxx/clk.c create mode 100644 cpu/at32uc/at32uc3a0xxx/portmux.c create mode 100644 cpu/at32uc/at32uc3a0xxx/sm.h create mode 100644 cpu/at32uc/cache.c create mode 100644 cpu/at32uc/config.mk create mode 100644 cpu/at32uc/cpu.c create mode 100644 cpu/at32uc/exception.c create mode 100644 cpu/at32uc/flashc.c create mode 100644 cpu/at32uc/flashc.h create mode 100644 cpu/at32uc/interrupts.c create mode 100644 cpu/at32uc/portmux-gpio.c create mode 100644 cpu/at32uc/sdramc.c create mode 100644 cpu/at32uc/sdramc.h create mode 100644 cpu/at32uc/smc.c create mode 100644 cpu/at32uc/smc.h create mode 100644 cpu/at32uc/start.S create mode 100644 include/asm-avr32/arch-at32uc3a0xxx/addrspace.h create mode 100644 include/asm-avr32/arch-at32uc3a0xxx/cacheflush.h create mode 100644 include/asm-avr32/arch-at32uc3a0xxx/chip-features.h create mode 100644 include/asm-avr32/arch-at32uc3a0xxx/clk.h create mode 100644 include/asm-avr32/arch-at32uc3a0xxx/gpio-impl.h create mode 100644 include/asm-avr32/arch-at32uc3a0xxx/gpio.h create mode 100644 include/asm-avr32/arch-at32uc3a0xxx/hmatrix.h create mode 100644 include/asm-avr32/arch-at32uc3a0xxx/memory-map.h create mode 100644 include/asm-avr32/arch-at32uc3a0xxx/portmux.h create mode 100644 include/asm-avr32/sram.h
cpu/at32uc/Makefile new file +57 -0
@@ -0,0 +1,57 @@ +# +# (C) Copyright 2000-2006 +# Wolfgang Denk, DENX Software Engineering, wd at denx.de. +# +# Copyright (C) 2005-2006 Atmel Corporation. +# +# See file CREDITS for list of people who contributed to this +# project. +# +# This program is free software; you can redistribute it and/or +# modify it under the terms of the GNU General Public License as +# published by the Free Software Foundation; either version 2 of +# the License, or (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program; if not, write to the Free Software +# Foundation, Inc., 59 Temple Place, Suite 330, Boston, +# MA 02111-1307 USA +# + +include $(TOPDIR)/config.mk + +LIB := $(obj)lib$(CPU).a + +START-y += start.o + +COBJS-y += cpu.o +COBJS-y += sdramc.o +COBJS-y += exception.o +COBJS-y += cache.o +COBJS-y += interrupts.o +COBJS-$(CONFIG_PORTMUX_GPIO) += portmux-gpio.o +COBJS-y += flashc.o +COBJS-y += smc.o + +SRCS := $(START-y:.o=.S) $(SOBJS-y:.o=.S) $(COBJS-y:.o=.c) +OBJS := $(addprefix $(obj),$(SOBJS-y) $(COBJS-y)) +START := $(addprefix $(obj),$(START-y)) + +all: $(obj).depend $(START) $(LIB) + +$(LIB): $(OBJS) + $(AR) $(ARFLAGS) $@ $^ + +######################################################################### + +# defines $(obj).depend target +include $(SRCTREE)/rules.mk + +sinclude $(obj).depend + +#########################################################################
cpu/at32uc/at32uc3a0xxx/Makefile new file +43 -0
43 lines not on this page · raw message
cpu/at32uc/at32uc3a0xxx/clk.c new file +88 -0
88 lines not on this page · raw message
cpu/at32uc/at32uc3a0xxx/portmux.c new file +154 -0
154 lines not on this page · raw message
cpu/at32uc/at32uc3a0xxx/sm.h new file +247 -0
247 lines not on this page · raw message
cpu/at32uc/cache.c new file +34 -0
@@ -0,0 +1,34 @@ +/* +* Copyright (C) 2004-2006 Atmel Corporation +* +* See file CREDITS for list of people who contributed to this +* project. +* +* This program is free software; you can redistribute it and/or +* modify it under the terms of the GNU General Public License as +* published by the Free Software Foundation; either version 2 of +* the License, or (at your option) any later version. +* +* This program is distributed in the hope that it will be useful, +* but WITHOUT ANY WARRANTY; without even the implied warranty of +* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +* GNU General Public License for more details. +* +* You should have received a copy of the GNU General Public License +* along with this program; if not, write to the Free Software +* Foundation, Inc., 59 Temple Place, Suite 330, Boston, +* MA 02111-1307 USA +*/ + +#include <common.h> + +/* +* This is called after loading something into memory. We need to +* make sure that everything that was loaded is actually written to +* RAM, and that the icache will look for it. Cleaning the dcache and +* invalidating the icache will do the trick. +*/ +void flush_cache (unsigned long start_addr,unsigned long size) +{ +/* No cache to clean in the at32uc3. */ +}
cpu/at32uc/config.mk new file +22 -0
@@ -0,0 +1,22 @@ +# +# Copyright (C) 2005-2006 Atmel Corporation +# +# See file CREDITS for list of people who contributed to this +# project. +# +# This program is free software; you can redistribute it and/or +# modify it under the terms of the GNU General Public License as +# published by the Free Software Foundation; either version 2 of +# the License, or (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program; if not, write to the Free Software +# Foundation, Inc., 59 Temple Place, Suite 330, Boston, +# MA 02111-1307 USA +# +PLATFORM_RELFLAGS += -mcpu=uc3a0512es
cpu/at32uc/cpu.c new file +79 -0
79 lines not on this page · raw message
cpu/at32uc/exception.c new file +121 -0
121 lines not on this page · raw message
cpu/at32uc/flashc.c new file +144 -0
144 lines not on this page · raw message
cpu/at32uc/flashc.h new file +84 -0
84 lines not on this page · raw message
cpu/at32uc/interrupts.c new file +150 -0
150 lines not on this page · raw message
cpu/at32uc/portmux-gpio.c new file +78 -0
78 lines not on this page · raw message
cpu/at32uc/sdramc.c new file +120 -0
120 lines not on this page · raw message
cpu/at32uc/sdramc.h new file +137 -0
137 lines not on this page · raw message
cpu/at32uc/smc.c new file +61 -0
61 lines not on this page · raw message
cpu/at32uc/smc.h new file +105 -0
105 lines not on this page · raw message
cpu/at32uc/start.S new file +252 -0
252 lines not on this page · raw message
include/asm-avr32/arch-at32uc3a0xxx/addrspace.h new file +63 -0
63 lines not on this page · raw message
include/asm-avr32/arch-at32uc3a0xxx/cacheflush.h new file +41 -0
41 lines not on this page · raw message
include/asm-avr32/arch-at32uc3a0xxx/chip-features.h new file +30 -0
@@ -0,0 +1,30 @@ +/* +* Copyright (C) 2007 Atmel Corporation +* +* See file CREDITS for list of people who contributed to this +* project. +* +* This program is free software; you can redistribute it and/or +* modify it under the terms of the GNU General Public License as +* published by the Free Software Foundation; either version 2 of +* the License, or (at your option) any later version. +* +* This program is distributed in the hope that it will be useful, +* but WITHOUT ANY WARRANTY; without even the implied warranty of +* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +* GNU General Public License for more details. +* +* You should have received a copy of the GNU General Public License +* along with this program; if not, write to the Free Software +* Foundation, Inc., 59 Temple Place, Suite 330, Boston, +* MA 02111-1307 USA +*/ +#ifndef __ASM_AVR32_ARCH_CHIP_FEATURES_H__ +#define __ASM_AVR32_ARCH_CHIP_FEATURES_H__ + +/* All the UC3A0xxx chips have these. */ +#define AT32UC3A0xxx_CHIP_HAS_USART +#define AT32UC3A0xxx_CHIP_HAS_SPI +#define AT32UC3A0xxx_CHIP_HAS_MACB + +#endif /* __ASM_AVR32_ARCH_CHIP_FEATURES_H__ */
include/asm-avr32/arch-at32uc3a0xxx/clk.h new file +85 -0
85 lines not on this page · raw message
include/asm-avr32/arch-at32uc3a0xxx/gpio-impl.h new file +62 -0
62 lines not on this page · raw message
include/asm-avr32/arch-at32uc3a0xxx/gpio.h new file +44 -0
44 lines not on this page · raw message
include/asm-avr32/arch-at32uc3a0xxx/hmatrix.h new file +47 -0
47 lines not on this page · raw message
include/asm-avr32/arch-at32uc3a0xxx/memory-map.h new file +75 -0
75 lines not on this page · raw message
include/asm-avr32/arch-at32uc3a0xxx/portmux.h new file +80 -0
80 lines not on this page · raw message
include/asm-avr32/sram.h new file +34 -0
@@ -0,0 +1,34 @@ +#ifndef __ASM_AVR32_SRAM_H +#define __ASM_AVR32_SRAM_H +#include <asm/types.h> + +struct sram_config { +/* Number of data bits. */ +u8 data_bits; + +/* Chip select */ +u8 chip_select; + +/* Number of address bits */ +u8 address_bits; + +/* nwe/nrd waveforms */ +u8 nwe_setup, nwe_pulse, nwe_hold; +u8 nrd_setup, nrd_pulse, nrd_hold; + +/* ncs waveforms */ +u8 ncs_wr_setup, ncs_wr_pulse, ncs_wr_hold; +u8 ncs_rd_setup, ncs_rd_pulse, ncs_rd_hold; + +/* Cycle length */ +u16 nwe_cycle, nrd_cycle; + +/* mode */ +u8 read_mode, write_mode, exnw_mode; +u8 bat, dbw, tdf_cycles, tdf_mode, pmen, ps; +}; + + +unsigned long sram_init (const struct sram_config *config); + +#endif /* __ASM_AVR32_SRAM_H */
Signature · 1 line
1.6.0.4
PATCH v2 9/9 AVR32: Board support for ATEVK1100
This patch adds support for the ATEVK1100 evaluation kit, which contains an AT32UC3A0512ES chip.
--- MAINTAINERS | 1 + MAKEALL | 1 + Makefile | 3 + board/atmel/atevk1100/Makefile | 44 ++++++++ board/atmel/atevk1100/atevk1100.c | 122 ++++++++++++++++++++++ board/atmel/atevk1100/config.mk | 4 + board/atmel/atevk1100/u-boot.lds | 73 +++++++++++++ include/configs/atevk1100.h | 207 +++++++++++++++++++++++++++++++++++++ 8 files changed, 455 insertions(+), 0 deletions(-) create mode 100644 board/atmel/atevk1100/Makefile create mode 100644 board/atmel/atevk1100/atevk1100.c create mode 100644 board/atmel/atevk1100/config.mk create mode 100644 board/atmel/atevk1100/u-boot.lds create mode 100644 include/configs/atevk1100.h
MAINTAINERS +1 -0
@@ -747,6 +747,7 @@ Haavard Skinnemoen <[email protected]> ATSTK1004 AT32AP7002 ATSTK1006 AT32AP7000 ATNGW100 AT32AP7000 + ATEVK1100 AT32UC3A0512 ######################################################################### # SuperH Systems: #
MAKEALL +1 -0
@@ -720,6 +720,7 @@ LIST_coldfire=" \ ######################################################################### LIST_avr32=" \ + atevk1100 \ atstk1002 \ atstk1003 \ atstk1004 \
Makefile +3 -0
@@ -3047,6 +3047,9 @@ $(BFIN_BOARDS): # AVR32 #======================================================================== +atevk1100_config : unconfig + @$(MKCONFIG) $(@:_config=) avr32 at32uc atevk1100 atmel at32uc3a0xxx + atngw100_config : unconfig @$(MKCONFIG) $(@:_config=) avr32 at32ap atngw100 atmel at32ap700x
board/atmel/atevk1100/Makefile new file +44 -0
44 lines not on this page · raw message
board/atmel/atevk1100/atevk1100.c new file +122 -0
122 lines not on this page · raw message
board/atmel/atevk1100/config.mk new file +4 -0
@@ -0,0 +1,4 @@ +PLATFORM_RELFLAGS += -ffunction-sections -fdata-sections +PLATFORM_LDFLAGS += --gc-sections +TEXT_BASE = 0x80000000 +LDSCRIPT = $(obj)board/atmel/atevk1100/u-boot.lds
board/atmel/atevk1100/u-boot.lds new file +73 -0
73 lines not on this page · raw message
include/configs/atevk1100.h new file +207 -0
207 lines not on this page · raw message
Signature · 1 line
1.6.0.4
PATCH v2 9/9 AVR32: Board support for ATEVK1100
Dear Gunnar Rangoy,
Quoted 2 lines · In message <7d98d3ebed68dfca910943a360d0b8aaa28d9c17.1232710611.git.gunnar@rang…
This patch adds support for the ATEVK1100 evaluation kit, which contains an AT32UC3A0512ES chip.
...
...
Quoted 2 lines
board/atmel/atevk1100/atevk1100.c new file +0
-0
...
board/atmel/atevk1100/atevk1100.c line 52
Quoted 8 lines
board/atmel/atevk1100/atevk1100.c new file +8
-0+/* +.ncs_wr_pulse = 1, +.ncs_rd_pulse = 1, +.nwe_pulse = 1, +.nrd_pulse = 1, +.nwe_cycle = 2, +.nrd_cycle = 2, +*/
All this is dead code. Please remove.
...
board/atmel/atevk1100/atevk1100.c
Quoted 9 lines
board/atmel/atevk1100/atevk1100.c new file +9
-0+/* +* Command line configuration. +*/ +#include <config_cmd_default.h> + +#define CONFIG_CMD_ASKENV +#define CONFIG_CMD_DHCP + +#undef CONFIG_CMD_AUTOSCRIPT
Is there a specific reason for not allowing to use shell scripts?
Best regards,
Wolfgang Denk
Signature · 5 lines
DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
I'm a programmer: I don't buy software, I write it.
-- Tom ChristiansenPATCH v2 5/9 AVR32: macb - Disable 100mbps if clock is slow
In reply to Gunnar RangoyQuoted 18 lines · Gunnar Rangoy
From: Olav Morken <[email protected]>
For 100mbps operation, the ethernet controller requires a 25 MHz clock in MII mode, and a 50 MHz clock in RMII mode. If the clock is slower, disable 100 Mbps mode.
Signed-off-by Gunnar Rangoy <[email protected]>Signed-off-by Paul Driveklepp <[email protected]>Signed-off-by Olav Morken <[email protected]>--- drivers/net/macb.c | 12 +++++++++++- 1 files changed, 11 insertions(+), 1 deletions(-)drivers/net/macb.c +0
-0
drivers/net/macb.c static void macb_phy_reset(struct macb_device *macb) · line 300
Quoted 5 lines
drivers/net/macb.c +1
-1@@ -297,7 +297,17 @@ static void macb_phy_reset(struct macb_device *macb)int i;u16 status, adv; - adv =ADVERTISE_CSMA |ADVERTISE_ALL ; + adv =ADVERTISE_CSMA |ADVERTISE_ALL ;
??
drivers/net/macb.c static void macb_phy_reset(struct macb_device *macb) · line 309
Quoted 9 lines
drivers/net/macb.c +9
-0+ +#ifdef CONFIG_MACB_FORCE10M +printf ("%s: 100Mbps is not supported on this board - forcing 10Mbps.\n" , + netdev->name); + + adv &= ~ADVERTISE_100FULL ; + adv &= ~ADVERTISE_100HALF ; + adv &= ~ADVERTISE_100BASE4 ; +#endif
not a fan
could you be more specific about the problem?
Best Regards,
J.
PATCH v2 8/9 AVR32: CPU support for AT32UC3A0xxx CPUs
In reply to Gunnar RangoyDear Gunnar Rangoy,
Quoted 1 line · In message <b47b53bb44d0d937f9a40249c1cd668640aec400.1232710611.git.gunnar@rang…
This patch adds support for the AT32UC3A0xxx chips.
...
Quoted 2 lines
+++
b/cpu/at32uc/at32uc3a0xxx/portmux.c(quoted diff) +0
-0@@ -0,0 +1,154 @@
...
cpu/at32uc/at32uc3a0xxx/portmux.c line 62
Quoted 13 lines
cpu/at32uc/at32uc3a0xxx/portmux.c new file +13
-0+portmux_select_peripheral (PORTMUX_PORT (0 ), +0x0003C000 | +0x1E000000 ,PORTMUX_FUNC_C ,0 ); +portmux_select_peripheral (PORTMUX_PORT (1 ), +0x00000010 | +0x00007C00 | +0x00030000 | +0xE0000000 ,PORTMUX_FUNC_C ,0 ); +portmux_select_peripheral (PORTMUX_PORT (2 ), +0xFFFFFFC0 ,PORTMUX_FUNC_A ,0 ); +portmux_select_peripheral (PORTMUX_PORT (3 ), +0x00003FFF ,PORTMUX_FUNC_A ,0 ); +}
It would be nice if you used readable names instead of all these magic hardcoded constants.
Quoted 10 lines
+++
b/cpu/at32uc/at32uc3a0xxx/sm.hcpu/at32uc/at32uc3a0xxx/sm.h new file +8
-0@@ -0,0 +1,247 @@ +/* +* Register definitions for System Manager +*/ +#ifndef __CPU_AT32UC_SM_H__ +#define __CPU_AT32UC_SM_H__ + +/* SM register offsets */ +/* PM starts at 0xFFFF0C00 */
cpu/at32uc/at32uc3a0xxx/sm.h line 18
Quoted 10 lines
cpu/at32uc/at32uc3a0xxx/sm.h new file +10
-0+#define SM_PM_REGS_OFFSET 0x0c00 +#define SM_PM_MCCTRL (SM_PM_REGS_OFFSET + 0x0000) +#define SM_PM_CKSEL (SM_PM_REGS_OFFSET + 0x0004) +#define SM_PM_CPU_MASK (SM_PM_REGS_OFFSET + 0x0008) +#define SM_PM_HSB_MASK (SM_PM_REGS_OFFSET + 0x000c) +#define SM_PM_PBA_MASK (SM_PM_REGS_OFFSET + 0x0010) +#define SM_PM_PBB_MASK (SM_PM_REGS_OFFSET + 0x0014) +#define SM_PM_PLL0 (SM_PM_REGS_OFFSET + 0x0020) +#define SM_PM_PLL1 (SM_PM_REGS_OFFSET + 0x0024) +#define SM_PM_OSCCTRL0 (SM_PM_REGS_OFFSET + 0x0028)
...
Instead of using offsets everywhere I would appreciate if the code could be changed to use C structs instead (like we do in PPC land).
cpu/at32uc/cpu.c line 79
Quoted 11 lines
cpu/at32uc/cpu.c new file +11
-0+int do_reset (cmd_tbl_t *cmdtp,int flag,int argc,char *argv[]) +{ +/* This will reset the CPU core, caches, MMU and all internal busses */ +__builtin_mtdr (8 ,1 <<13 );/* set DC:DBE */ +__builtin_mtdr (8 ,1 <<30 );/* set DC:RES */ + +/* Flush the pipeline before we declare it a failure */ + asmvolatile ("sub pc, pc, -4" ); + +return -1 ; +}
I read this as if you just reset the CPU "internal" stuff. Sorry for asking stupid questions, I don't know this architecture at all, but:
Will external chips be reset this way, too? Or how do you make sure that external peripherals get properly reset?
Best regards,
Wolfgang Denk
Signature · 4 lines
DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de 365 Days of drinking Lo-Cal beer. = 1 Lite-year
PATCH v2 5/9 AVR32: macb - Disable 100mbps if clock is slow
In reply to Jean-Christophe PLAGNIOL-VILLARDQuoted 22 lines · Jean-Christophe PLAGNIOL-VILLARD
Quoted 21 lines · Gunnar Rangoy
From: Olav Morken <[email protected]>
For 100mbps operation, the ethernet controller requires a 25 MHz clock in MII mode, and a 50 MHz clock in RMII mode. If the clock is slower, disable 100 Mbps mode.
Signed-off-by Gunnar Rangoy <[email protected]>Signed-off-by Paul Driveklepp <[email protected]>Signed-off-by Olav Morken <[email protected]>--- drivers/net/macb.c | 12 +++++++++++- 1 files changed, 11 insertions(+), 1 deletions(-)drivers/net/macb.c +0
-0@@ -297,7 +297,17 @@ static void macb_phy_reset(struct macb_device *macb)int i;u16 status, adv;
Quoted 3 lines
drivers/net/macb.c static void macb_phy_reset(struct macb_device *macb) · line 300
Quoted 2 lines
drivers/net/macb.c +1
-1- adv =ADVERTISE_CSMA |ADVERTISE_ALL ; + adv =ADVERTISE_CSMA |ADVERTISE_ALL ;??
Oops... Will fix this.
Quoted 12 lines
drivers/net/macb.c static void macb_phy_reset(struct macb_device *macb) · line 309
Quoted 9 lines
drivers/net/macb.c +9
-0+ +#ifdef CONFIG_MACB_FORCE10M +printf ("%s: 100Mbps is not supported on this board - forcing 10Mbps.\n" , + netdev->name); + + adv &= ~ADVERTISE_100FULL ; + adv &= ~ADVERTISE_100HALF ; + adv &= ~ADVERTISE_100BASE4 ; +#endif not a fan
could you be more specific about the problem?
Quoted 3 lines
Best Regards,
J.
On the EVK1100 board, the CPU (UC3A0512) is connected to the PHY via an RMII bus. This requires the CPU clock to be at least 50 MHz. Unfortunately, the chip on current EVK1100 boards may be unable to run at more than 50 MHz, and with the oscillator on the board, the closest frequency we can generate is 48 MHz.
This patch makes it possible to limit the macb to 10 MBit for this case. We are open for suggestions for other solutions.
Signature · 2 lines
Gunnar Rang?y, 99360699
PATCH v2 9/9 AVR32: Board support for ATEVK1100
In reply to Wolfgang DenkQuoted 20 lines · Wolfgang Denk
Dear Gunnar Rangoy,
Quoted 2 lines · In message <7d98d3ebed68dfca910943a360d0b8aaa28d9c17.1232710611.git.gunnar@rang…
This patch adds support for the ATEVK1100 evaluation kit, which contains an AT32UC3A0512ES chip.
...
...
Quoted 2 lines
board/atmel/atevk1100/atevk1100.c new file +0
-0...
board/atmel/atevk1100/atevk1100.c line 52
Quoted 8 lines
board/atmel/atevk1100/atevk1100.c new file +8
-0+/* +.ncs_wr_pulse = 1, +.ncs_rd_pulse = 1, +.nwe_pulse = 1, +.nrd_pulse = 1, +.nwe_cycle = 2, +.nrd_cycle = 2, +*/
Quoted 1 line
All this is dead code. Please remove.
Will do that.
Quoted 16 lines
...
board/atmel/atevk1100/atevk1100.c
Quoted 9 lines
board/atmel/atevk1100/atevk1100.c new file +9
-0+/* +* Command line configuration. +*/ +#include <config_cmd_default.h> + +#define CONFIG_CMD_ASKENV +#define CONFIG_CMD_DHCP + +#undef CONFIG_CMD_AUTOSCRIPTIs there a specific reason for not allowing to use shell scripts?
Best regards,
Quoted 9 lines
Wolfgang Denk
--
DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de I'm a programmer: I don't buy software, I write it.
-- Tom Christiansen
No, not really. We did this because this was done in other AVR32 boards we used as references.
Signature · 2 lines
Gunnar Rang?y, 99360699
PATCH v2 8/9 AVR32: CPU support for AT32UC3A0xxx CPUs
In reply to Wolfgang DenkQuoted 22 lines · Wolfgang Denk
Dear Gunnar Rangoy,
Quoted 1 line · In message <b47b53bb44d0d937f9a40249c1cd668640aec400.1232710611.git.gunnar@rang…
This patch adds support for the AT32UC3A0xxx chips.
...
Quoted 2 lines
+++
b/cpu/at32uc/at32uc3a0xxx/portmux.c(quoted diff) +0
-0@@ -0,0 +1,154 @@...
cpu/at32uc/at32uc3a0xxx/portmux.c line 62
Quoted 13 lines
cpu/at32uc/at32uc3a0xxx/portmux.c new file +13
-0+portmux_select_peripheral (PORTMUX_PORT (0 ), +0x0003C000 | +0x1E000000 ,PORTMUX_FUNC_C ,0 ); +portmux_select_peripheral (PORTMUX_PORT (1 ), +0x00000010 | +0x00007C00 | +0x00030000 | +0xE0000000 ,PORTMUX_FUNC_C ,0 ); +portmux_select_peripheral (PORTMUX_PORT (2 ), +0xFFFFFFC0 ,PORTMUX_FUNC_A ,0 ); +portmux_select_peripheral (PORTMUX_PORT (3 ), +0x00003FFF ,PORTMUX_FUNC_A ,0 ); +}
Quoted 2 lines
It would be nice if you used readable names instead of all these magic hardcoded constants.
Each bit in the masks represents a single pin on the microcontroller. The readable names would therefore become something like:
#define EBI_PINS_PORT0 (0x0003C000 | 0x1E000000)
#define EBI_PINS_PORT1 (0x00000010 | 0x00007C00 | ...)
I'm not certain that that would make it much more readable. The bitmasks are more or less based on reading the datasheet and counting which pins are used.
Maybe adding a comment before the code would be just as useful?
Quoted 22 lines
Quoted 20 lines
+++
b/cpu/at32uc/at32uc3a0xxx/sm.hcpu/at32uc/at32uc3a0xxx/sm.h new file +18
-0@@ -0,0 +1,247 @@ +/* +* Register definitions for System Manager +*/ +#ifndef __CPU_AT32UC_SM_H__ +#define __CPU_AT32UC_SM_H__ + +/* SM register offsets */ +/* PM starts at 0xFFFF0C00 */ +#define SM_PM_REGS_OFFSET 0x0c00 +#define SM_PM_MCCTRL (SM_PM_REGS_OFFSET + 0x0000) +#define SM_PM_CKSEL (SM_PM_REGS_OFFSET + 0x0004) +#define SM_PM_CPU_MASK (SM_PM_REGS_OFFSET + 0x0008) +#define SM_PM_HSB_MASK (SM_PM_REGS_OFFSET + 0x000c) +#define SM_PM_PBA_MASK (SM_PM_REGS_OFFSET + 0x0010) +#define SM_PM_PBB_MASK (SM_PM_REGS_OFFSET + 0x0014) +#define SM_PM_PLL0 (SM_PM_REGS_OFFSET + 0x0020) +#define SM_PM_PLL1 (SM_PM_REGS_OFFSET + 0x0024) +#define SM_PM_OSCCTRL0 (SM_PM_REGS_OFFSET + 0x0028)...
Quoted 2 lines
Instead of using offsets everywhere I would appreciate if the code could be changed to use C structs instead (like we do in PPC land).
It was done like this because it was done in the existing AVR32 code which we based our work on. Since we used parts of that code, it made most sense to use the same coding style.
We can change it if you think it is necessary.
Quoted 17 lines
cpu/at32uc/cpu.c line 79
Quoted 12 lines
cpu/at32uc/cpu.c new file +11
-0+int do_reset (cmd_tbl_t *cmdtp,int flag,int argc,char *argv[]) +{ +/* This will reset the CPU core, caches, MMU and all internal busses */ +__builtin_mtdr (8 ,1 <<13 );/* set DC:DBE */ +__builtin_mtdr (8 ,1 <<30 );/* set DC:RES */ + +/* Flush the pipeline before we declare it a failure */ + asmvolatile ("sub pc, pc, -4" ); + +return -1 ; +}I read this as if you just reset the CPU "internal" stuff. Sorry for asking stupid questions, I don't know this architecture at all, but:
Will external chips be reset this way, too? Or how do you make sure that external peripherals get properly reset?
As most of the needed functionality is embedded in the microcontroller, there are very few external peripherals used by U-Boot. Apart from external memory, and oscillator, and level-shifters for the serial-port, there is only the ethernet PHY, and that one shouldn't need a reset.
Gunnar Rang?y,
99360699
PATCH v2 9/9 AVR32: Board support for ATEVK1100
In reply to Gunnar RangøyDear ?ISO-8859-1?Q?Gunnar_Rang?y?,
In message <[email protected]> you wrote:
...
Quoted 3 lines
Quoted 3 lines
Quoted 1 line
+#undef
CONFIG_CMD_AUTOSCRIPTIs there a specific reason for not allowing to use shell scripts?
...
Quoted 2 lines
No, not really. We did this because this was done in other AVR32 boards we used as references.
Then pleasse don't - let's overcome the bad examples.
Best regards,
Wolfgang Denk
Signature · 4 lines
DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de Never put off until tomorrow what you can put off indefinitely.
PATCH v2 8/9 AVR32: CPU support for AT32UC3A0xxx CPUs
In reply to Gunnar RangøyDear =?ISO-8859-1?Q?Gunnar_Rang=F8y?=,
Quoted 19 lines · In message <[email protected]> you…
Quoted 4 lines
cpu/at32uc/at32uc3a0xxx/portmux.c
Quoted 4 lines
cpu/at32uc/at32uc3a0xxx/portmux.c new file +3
-0+portmux_select_peripheral (PORTMUX_PORT (0 ), +0x0003C000 | + 0x1E000000, PORTMUX_FUNC_C=, 0);
Quoted 5 lines
cpu/at32uc/at32uc3a0xxx/portmux.c
Quoted 5 lines
cpu/at32uc/at32uc3a0xxx/portmux.c new file +5
-0+portmux_select_peripheral (PORTMUX_PORT (1 ), +0x00000010 | +0x00007C00 | +0x00030000 | + 0xE0000000, PORTMUX_FUNC_C=, 0);
Quoted 2 lines
Review
Quoted 2 lines
(quoted diff) +2
-0+ portmux_select_peripheral(PORTMUX_PORT(2), + 0xFFFFFFC0, PORTMUX_FUNC_A=, 0);
Quoted 2 lines
Review
Quoted 2 lines
(quoted diff) +2
-0+ portmux_select_peripheral(PORTMUX_PORT(3), + 0x00003FFF, PORTMUX_FUNC_A=, 0);
Quoted 2 lines
Quoted 1 line
+}
Quoted 7 lines
Quoted 2 lines
It would be nice if you used readable names instead of all these magic hardcoded constants.
Each bit in the masks represents a single pin on the microcontroller. The readable names would therefore become something like:
#defineEBI_PINS_PORT0(0x0003C000|0x1E000000)
#defineEBI_PINS_PORT1(0x00000010|0x00007C00| ...)
Um, no. You didn't get it. You use magic numbers again.
That should be some
#define FOO_PIN_XXX 0x0003C000
#define FOO_PIN_YYY 0x1E000000
#define BAR_PIN_XXX 0x00000010
#DEFINE BAR_PIN_YYY 0x00007C00
in one place and
#define EBI_PINS_PORT0 (FOO_PIN_XXX | FOO_PIN_YYY) #define EBI_PINS_PORT1 ((BAR_PIN_XXX| BAR_PIN_YYY)
elsewhere, using readable names for the defines.
Quoted 3 lines
I'm not certain that that would make it much more readable. The bitmasks are more or less based on reading the datasheet and counting which pins are used.
This is exactly what should NOT be necessary to read and understand the code.
If I see a line like
rtx->txbd[txIdx].cbd_sc |= (BD_ENET_TX_READY | BD_ENET_TX_LAST |BD_ENET_TX_WRAP);
somewhere in a driver (obviously dealing with ethernet), I don;t have to look up any specific bits or ping numbers - I can just read it.
If somebody else writes the (computationally equivalent) code
rtx->txbd[txIdx].cbd_sc |= 0xA800; or even rtx->txbd[txIdx].cbd_sc |= (0x8000 | 0x0800 | 0x2000);
then I don;t have a clue what's going on, and it takes me precious time to look this up - me and everybody else, each and every time again when I'm reading the code.
This is simply not acceptable.
Quoted 1 line
Maybe adding a comment before the code would be just as useful?
No, a comment can't fix this, especially as you just have to wait a few months to see the comment and the code getting out of sync.
Please fix the code.
Quoted 8 lines
Quoted 2 lines
Instead of using offsets everywhere I would appreciate if the code could be changed to use C structs instead (like we do in PPC land).
It was done like this because it was done in the existing AVR32 code which we based our work on. Since we used parts of that code, it made most sense to use the same coding style.
We can change it if you think it is necessary.
Yes, please change it. And patches to fix the existing code that doe sthis are welcome, too.
Quoted 18 lines
Quoted 3 lines
cpu/at32uc/cpu.c
Quoted 3 lines
cpu/at32uc/cpu.c new file +3
-0+int do_reset (cmd_tbl_t *cmdtp,int flag,int argc,char *argv[]) +{ +/* This will reset the CPU core, caches, MMU and all internal buss= es */
Quoted 5 lines
Quoted 5 lines
cpu/at32uc/cpu.c new file +5
-0+__builtin_mtdr (8 ,1 <<13 );/* set DC:DBE */ +__builtin_mtdr (8 ,1 <<30 );/* set DC:RES */ + +/* Flush the pipeline before we declare it a failure */ + asmvolatile ("sub pc, pc, -4" );Quoted 8 lines
cpu/at32uc/cpu.c line 79
Quoted 3 lines
cpu/at32uc/cpu.c new file +3
-0+ +return -1 ; +}I read this as if you just reset the CPU "internal" stuff. Sorry for asking stupid questions, I don't know this architecture at all, but:
Will external chips be reset this way, too? Or how do you make sure that external peripherals get properly reset?
Quoted 4 lines
As most of the needed functionality is embedded in the microcontroller, there are very few external peripherals used by U-Boot. Apart from external memory, and oscillator, and level-shifters for the serial-port, there is only the ethernet PHY, and that one shouldn't need a reset.
Famous last words. What if exactrly the PHY is stuck and needs a reset?
Hmmm... "apart from external memory" ... does externam memory also include NOR flash? Eventually the NOR flash you are booting from? Assume the NOR flash is in query mode when you reset the board - how does it get reset, then?
Best regards,
Wolfgang Denk
Signature · 6 lines
DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
It is impractical for the standard to attempt to constrain the
behavior of code that does not obey the constraints of the standard.
- Doug GwynPATCH v2 9/9 AVR32: Board support for ATEVK1100
In reply to Wolfgang DenkQuoted 4 lines · Wolfgang Denk
Dear ?ISO-8859-1?Q?Gunnar_Rang?y?,
In message <[email protected]> you wrote:
...
Quoted 9 lines
Quoted 3 lines
Quoted 3 lines
Quoted 1 line
+#undef
CONFIG_CMD_AUTOSCRIPTIs there a specific reason for not allowing to use shell scripts?
...
Quoted 2 lines
No, not really. We did this because this was done in other AVR32 boards we used as references.
Then pleasse don't - let's overcome the bad examples.
OK, this is fixed.
Regards Gunnar
PATCH v2 5/9 AVR32: macb - Disable 100mbps if clock is slow
In reply to Gunnar RangøyHi Gunnar,
Quoted 53 lines · Gunnar Rang?y
Quoted 23 lines · Jean-Christophe PLAGNIOL-VILLARD
Quoted 21 lines · Gunnar Rangoy
From: Olav Morken <[email protected]>
For 100mbps operation, the ethernet controller requires a 25 MHz clock in MII mode, and a 50 MHz clock in RMII mode. If the clock is slower, disable 100 Mbps mode.
Signed-off-by Gunnar Rangoy <[email protected]>Signed-off-by Paul Driveklepp <[email protected]>Signed-off-by Olav Morken <[email protected]>--- drivers/net/macb.c | 12 +++++++++++- 1 files changed, 11 insertions(+), 1 deletions(-)drivers/net/macb.c +0
-0@@ -297,7 +297,17 @@ static void macb_phy_reset(struct macb_device *macb)int i;u16 status, adv;Quoted 5 lines
drivers/net/macb.c static void macb_phy_reset(struct macb_device *macb)
Quoted 3 lines
drivers/net/macb.c +1
-1- adv =ADVERTISE_CSMA |ADVERTISE_ALL ; + adv =ADVERTISE_CSMA |ADVERTISE_ALL ;??
Oops... Will fix this.
Quoted 13 lines
drivers/net/macb.c static void macb_phy_reset(struct macb_device *macb)
Quoted 10 lines
drivers/net/macb.c +9
-0+ +#ifdef CONFIG_MACB_FORCE10M +printf ("%s: 100Mbps is not supported on this board - forcing 10Mbps.\n" , + netdev->name); + + adv &= ~ADVERTISE_100FULL ; + adv &= ~ADVERTISE_100HALF ; + adv &= ~ADVERTISE_100BASE4 ; +#endif not a fan
could you be more specific about the problem?Quoted 4 lines
Best Regards,
J.
Quoted 10 lines
On the EVK1100 board, the CPU (UC3A0512) is connected to the PHY via an RMII bus. This requires the CPU clock to be at least 50 MHz. Unfortunately, the chip on current EVK1100 boards may be unable to run at more than 50 MHz, and with the oscillator on the board, the closest frequency we can generate is 48 MHz.
This patch makes it possible to limit the macb to 10 MBit for this case. We are open for suggestions for other solutions.
How about using a PHY capability override CONFIG. Something like this:
#if defined(CONFIG_MACB_PHY_CAPAB) <-- insert better name here
adv = ADVERTISE_CSMA | CONFIG_MACB_PHY_CAPAB;
#else
adv = ADVERTISE_CSMA | ADVERTISE_ALL
#endif
Just an idea...
regards,
Ben
PATCH v2 1/9 Fix IP alignement problem
In reply to Gunnar RangoyQuoted 22 lines · Gunnar Rangoy
From: Olav Morken <[email protected]>
This patch removes volatile from:
volatileIP_t*ip = (IP_t*)xip;Due to a bug, avr32-gcc will assume that ip is aligned on a word boundary when using volatile, which causes an exception since xip isn't aligned on a word boundary.
Signed-off-by Gunnar Rangoy <[email protected]>Signed-off-by Paul Driveklepp <[email protected]>Signed-off-by Olav Morken <[email protected]>--- net/net.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)net/net.c +0
-0
net/net.c NetSetEther(volatile uchar * xet, uchar * addr, uint prot) · line 1691
Quoted 8 lines
net/net.c +1
-1@@ -1685,7 +1685,7 @@ NetSetEther(volatile uchar * xet, uchar * addr, uint prot)void NetSetIP (volatile uchar * xip,IPaddr_t dest,int dport,int sport,int len) { -volatile IP_t *ip = (IP_t *)xip; +IP_t *ip = (IP_t *)xip;/* * If the data is an odd number of bytes, zero the
This seems reasonable. I'll try it out and will get back to you soon.
regards,
Ben
PATCH v2 5/9 AVR32: macb - Disable 100mbps if clock is slow
In reply to Gunnar RangøyQuoted 16 lines
Quoted 12 lines
drivers/net/macb.c static void macb_phy_reset(struct macb_device *macb) · line 309
Quoted 9 lines
drivers/net/macb.c +9
-0+ +#ifdef CONFIG_MACB_FORCE10M +printf ("%s: 100Mbps is not supported on this board - forcing 10Mbps.\n" , + netdev->name); + + adv &= ~ADVERTISE_100FULL ; + adv &= ~ADVERTISE_100HALF ; + adv &= ~ADVERTISE_100BASE4 ; +#endif not a fan
could you be more specific about the problem?Quoted 3 lines
Best Regards,
J.
Quoted 5 lines
On the EVK1100 board, the CPU (UC3A0512) is connected to the PHY via an RMII bus. This requires the CPU clock to be at least 50 MHz. Unfortunately, the chip on current EVK1100 boards may be unable to run at more than 50 MHz, and with the oscillator on the board, the closest frequency we can generate is 48 MHz.
IMHO It's a HW design error to not use the MII
Quoted 3 lines
This patch makes it possible to limit the macb to 10 MBit for this case. We are open for suggestions for other solutions.
I guest you may need to disable the phy auto config mode and force him to be see as a 10Mbps phy evenif it's a 10/100
Best Regards,
J.
PATCH v2 5/9 AVR32: macb - Disable 100mbps if clock is slow
Quoted 6 lines · Jean-Christophe PLAGNIOL-VILLARD
Quoted 5 lines
On the EVK1100 board, the CPU (UC3A0512) is connected to the PHY via an RMII bus. This requires the CPU clock to be at least 50 MHz. Unfortunately, the chip on current EVK1100 boards may be unable to run at more than 50 MHz, and with the oscillator on the board, the closest frequency we can generate is 48 MHz.
IMHO It's a HW design error to not use the MII
Some people want to use the extra pins for other things...
Unfortunately, there are quite a few boards with early engineering samples around, and they have various issues. The chips that are in production are capable of running fast enough to support RMII.
Quoted 4 lines
Quoted 2 lines
This patch makes it possible to limit the macb to 10 MBit for this case. We are open for suggestions for other solutions.
I guest you may need to disable the phy auto config mode and force him to be see as a 10Mbps phy evenif it's a 10/100
No need to disable autonegotiation -- you still want to select between half an full duplex, for example. But you'll need to limit the available options to the ones that actually work.
That said, I kind of like Ben's suggestion -- it's a more general solution to all sorts of board/phy/chip/whatever limitations.
As for a better name, how about CONFIG_MACB_ADVERTISE?
Haavard
PATCH v2 5/9 AVR32: macb - Disable 100mbps if clock is slow
Quoted 9 lines · Haavard Skinnemoen
Quoted 6 lines · Jean-Christophe PLAGNIOL-VILLARD
Quoted 5 lines
On the EVK1100 board, the CPU (UC3A0512) is connected to the PHY via an RMII bus. This requires the CPU clock to be at least 50 MHz. Unfortunately, the chip on current EVK1100 boards may be unable to run at more than 50 MHz, and with the oscillator on the board, the closest frequency we can generate is 48 MHz.
IMHO It's a HW design error to not use the MII
Some people want to use the extra pins for other things...
make sense
so I'll put a 10Mpbs phy personnaly instead or a 10/100 that can be put in a 10 mode instead to avoid to manage it in the code
Quoted 5 lines
Unfortunately, there are quite a few boards with early engineering samples around, and they have various issues. The chips that are in production are capable of running fast enough to support RMII.
Quoted 8 lines
Quoted 4 lines
Quoted 2 lines
This patch makes it possible to limit the macb to 10 MBit for this case. We are open for suggestions for other solutions.
I guest you may need to disable the phy auto config mode and force him to be see as a 10Mbps phy evenif it's a 10/100
No need to disable autonegotiation -- you still want to select between half an full duplex, for example. But you'll need to limit the available options to the ones that actually work.
I do not mean the autoneg but to specify to the phy, if possible, to never accept the 100Mps instead of do it in the code
Quoted 5 lines
That said, I kind of like Ben's suggestion -- it's a more general solution to all sorts of board/phy/chip/whatever limitations.
As for a better name, how about
CONFIG_MACB_ADVERTISE?
why not
Best Regards,
J.
PATCH v2 5/9 AVR32: macb - Disable 100mbps if clock is slow
In reply to Ben WarrenQuoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 7 lines
Quoted 4 lines
This patch makes it possible to limit the macb to 10 MBit for this case. We are open for suggestions for other solutions.
I guest you may need to disable the phy auto config mode and force him to be see as a 10Mbps phy evenif it's a 10/100
No need to disable autonegotiation -- you still want to select between half an full duplex, for example. But you'll need to limit the available options to the ones that actually work.
Quoted 6 lines
Quoted 3 lines
I do not mean the autoneg but to specify to the phy, if possible, to never accept the 100Mps instead of do it in the code
That's basically why the advertise register is r/w, so you can limit autonegotiation through software and don't have to allocate pins for strapping.
I've some phy/switch here that have on option to put them in 10M mode without have to manage it after even via the adverstise
I've just guess maybe the one on the board could allow it too
Quoted 3 lines
I like it too. One of the common checkbox items, though: do any Atmel chips have more than one MACB, in which case this should be
CONFIG_MACBx_ADVERTISEor something like that?
on at91 actually not
Best Regards,
J.
PATCH v2 5/9 AVR32: macb - Disable 100mbps if clock is slow
In reply to Jean-Christophe PLAGNIOL-VILLARDQuoted 10 lines · Jean-Christophe PLAGNIOL-VILLARD
Quoted 8 lines · Haavard Skinnemoen
Quoted 6 lines · Jean-Christophe PLAGNIOL-VILLARD
Quoted 6 lines
On the EVK1100 board, the CPU (UC3A0512) is connected to the PHY via an RMII bus. This requires the CPU clock to be at least 50 MHz. Unfortunately, the chip on current EVK1100 boards may be unable to run at more than 50 MHz, and with the oscillator on the board, the closest frequency we can generate is 48 MHz.
Quoted 8 lines
Quoted 4 lines
Quoted 2 lines
IMHO It's a HW design error to not use the MII
Some people want to use the extra pins for other things...
make sense
so I'll put a 10Mpbs phy personnaly instead or a 10/100 that can be put in a 10 mode instead to avoid to manage it in the code
I haven't shopped for PHYs in a while, but imagine it's probably hard to find one that's 10Mb only, and if so it's probably more expensive than 10/100
Quoted 11 lines
Quoted 11 lines
Unfortunately, there are quite a few boards with early engineering samples around, and they have various issues. The chips that are in production are capable of running fast enough to support RMII.
Quoted 6 lines
Quoted 3 lines
This patch makes it possible to limit the macb to 10 MBit for this case. We are open for suggestions for other solutions.
I guest you may need to disable the phy auto config mode and force him to be see as a 10Mbps phy evenif it's a 10/100
Quoted 7 lines
Quoted 4 lines
No need to disable autonegotiation -- you still want to select between half an full duplex, for example. But you'll need to limit the available options to the ones that actually work.
I do not mean the autoneg but to specify to the phy, if possible, to never accept the 100Mps instead of do it in the code
That's basically why the advertise register is r/w, so you can limit autonegotiation through software and don't have to allocate pins for strapping.
Quoted 7 lines
Quoted 5 lines
That said, I kind of like Ben's suggestion -- it's a more general solution to all sorts of board/phy/chip/whatever limitations.
As for a better name, how about
CONFIG_MACB_ADVERTISE?why not
I like it too. One of the common checkbox items, though: do any Atmel chips have more than one MACB, in which case this should be CONFIG_MACBx_ADVERTISE or something like that?
Quoted 3 lines
Best Regards,
J.
regards,
Ben
PATCH v2 5/9 AVR32: macb - Disable 100mbps if clock is slow
Quoted 3 lines · On Wed, 28 Jan 2009 15:40:49 -0800 Ben Warren <[email protected]> wrote:
Quoted 2 lines · Jean-Christophe PLAGNIOL-VILLARD
On 22:42 Wed 28 Jan , Haavard Skinnemoen wrote:
<snipp>
Quoted 8 lines
Quoted 4 lines
Quoted 2 lines
As for a better name, how about
CONFIG_MACB_ADVERTISE?why not
I like it too. One of the common checkbox items, though: do any Atmel chips have more than one MACB, in which case this should be
CONFIG_MACBx_ADVERTISEor something like that?
Yes, AP7000 have two Ethernet MACs. And if I got this right you want to make a generic config about it, so then I guess it should open up for having more than one MAC.
Signature · 2 lines
Best regards, Hans-Christian Egtvedt
Ben Warren
AppliedPATCH v2 1/9 Fix IP alignement problem
In reply to Gunnar RangoyQuoted 22 lines · Gunnar Rangoy
From: Olav Morken <[email protected]>
This patch removes volatile from:
volatileIP_t*ip = (IP_t*)xip;Due to a bug, avr32-gcc will assume that ip is aligned on a word boundary when using volatile, which causes an exception since xip isn't aligned on a word boundary.
Signed-off-by Gunnar Rangoy <[email protected]>Signed-off-by Paul Driveklepp <[email protected]>Signed-off-by Olav Morken <[email protected]>--- net/net.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)net/net.c +0
-0
net/net.c NetSetEther(volatile uchar * xet, uchar * addr, uint prot) · line 1691
Quoted 8 lines
net/net.c +1
-1@@ -1685,7 +1685,7 @@ NetSetEther(volatile uchar * xet, uchar * addr, uint prot)void NetSetIP (volatile uchar * xip,IPaddr_t dest,int dport,int sport,int len) { -volatile IP_t *ip = (IP_t *)xip; +IP_t *ip = (IP_t *)xip;/* * If the data is an odd number of bytes, zero the
Applied to net repo (fixed typo in title )
thanks,
Ben
PATCH v2 6/9 AVR32: macb - Search for PHY id
In reply to Gunnar RangoyQuoted 54 lines · Gunnar Rangoy
This patch adds support for searching through available PHY-addresses in the macb-driver. This is needed for the ATEVK1100 evaluation board, where the PHY-address will be initialized to either 1 or 7.
This patch adds a config option,
CONFIG_MACB_SEARCH_PHY, which when enabled tells the driver to search for the PHY address.Signed-off-by Gunnar Rangoy <[email protected]>Signed-off-by Paul Driveklepp <[email protected]>Signed-off-by Olav Morken <[email protected]>--- drivers/net/macb.c | 31 +++++++++++++++++++++++++++++++ 1 files changed, 31 insertions(+), 0 deletions(-)drivers/net/macb.c +24
-0@@ -327,6 +327,30 @@ static void macb_phy_reset(struct macb_device *macb) netdev->name, status); } +#ifdef CONFIG_MACB_SEARCH_PHY +static int macb_phy_find (struct macb_device *macb) +{ +int i; +u16 phy_id; + +/* Search for PHY... */ +for (i =0 ; i <32 ; i++) { + macb->phy_addr = i; + phy_id =macb_mdio_read (macb,MII_PHYSID1 ); +if (phy_id !=0xffff ) { +printf ("%s: PHY present at %d\n" , macb->netdev.name, i); +return 1 ; + } + } + +/* PHY isn't up to snuff */ +printf ("%s: PHY not found" , macb->netdev.name); + +return 0 ; +} +#endif /* CONFIG_MACB_SEARCH_PHY */ + +static int macb_phy_init (struct macb_device *macb) {struct eth_device *netdev = &macb->netdev;@@ -335,6 +359,13 @@ static int macb_phy_init(struct macb_device *macb)int media, speed, duplex;int i;
drivers/net/macb.c static int macb_phy_init(struct macb_device *macb) · line 371
Quoted 10 lines
drivers/net/macb.c +7
-0+#ifdef CONFIG_MACB_SEARCH_PHY +/* Auto-detect phy_addr */ +if (!macb_phy_find (macb)) { +return 0 ; + } +#endif /* CONFIG_MACB_SEARCH_PHY */ +/* Check if the PHY is up to snuff... */ phy_id =macb_mdio_read (macb,MII_PHYSID1 );if (phy_id ==0xffff ) {
Added to net repo.
thanks,
Ben
PATCH v2 5/9 AVR32: macb - Disable 100mbps if clock is slow
In reply to Ben WarrenQuoted 8 lines · Ben Warren
Quoted 4 lines · Jean-Christophe PLAGNIOL-VILLARD
make sense
so I'll put a 10Mpbs phy personnaly instead or a 10/100 that can be put in a 10 mode instead to avoid to manage it in the codeI haven't shopped for PHYs in a while, but imagine it's probably hard to find one that's 10Mb only, and if so it's probably more expensive than 10/100
Especially when the board really is _supposed_ to do 100Mbps just fine.
Quoted 10 lines
Quoted 7 lines
Quoted 4 lines
No need to disable autonegotiation -- you still want to select between half an full duplex, for example. But you'll need to limit the available options to the ones that actually work.
I do not mean the autoneg but to specify to the phy, if possible, to never accept the 100Mps instead of do it in the code
That's basically why the advertise register is r/w, so you can limit autonegotiation through software and don't have to allocate pins for strapping.
Yeah, I thought the advertise register was the correct way of informing the phy about what modes you support...
I suppose we could move the initialization of the advertise register to board code, but that's going to be a lot uglier since the board code would have to access the MACB hardware to drive the MDIO pins.
Perhaps some boards don't need any initialization at all, but then again, isn't it safer to unconditionally write the correct value? And when we're going to write the register _anyway_, we might as well allow the board code to influence the value we're going to write.
The more I think of this, the more I'm convinced we should have allowed that from the beginning. The current driver is making assumptions it shouldn't be making.
Quoted 10 lines
Quoted 7 lines
Quoted 5 lines
That said, I kind of like Ben's suggestion -- it's a more general solution to all sorts of board/phy/chip/whatever limitations.
As for a better name, how about
CONFIG_MACB_ADVERTISE?why not
I like it too. One of the common checkbox items, though: do any Atmel chips have more than one MACB, in which case this should be
CONFIG_MACBx_ADVERTISEor something like that?
Yes, AT32AP7000 has two MACBs, so that's probably a good idea.
Better yet, make it CONFIG_MACB_ADVERTISE(macb) so that the board can distinguish between various instances if it has to, and the driver doesn't have to do a long sequence of #ifdefs to find the correct value.
Haavard
PATCH v2 5/9 AVR32: macb - Disable 100mbps if clock is slow
In reply to Hans-Christian EgtvedtQuoted 13 lines · Hans-Christian Egtvedt
Quoted 3 lines · On Wed, 28 Jan 2009 15:40:49 -0800 Ben Warren <[email protected]> wrote:
Quoted 2 lines · Jean-Christophe PLAGNIOL-VILLARD
On 22:42 Wed 28 Jan , Haavard Skinnemoen wrote:
<snipp>
Quoted 4 lines
Quoted 4 lines
Quoted 2 lines
As for a better name, how about
CONFIG_MACB_ADVERTISE?why not
Quoted 8 lines
Quoted 4 lines
I like it too. One of the common checkbox items, though: do any Atmel chips have more than one MACB, in which case this should be
CONFIG_MACBx_ADVERTISEor something like that?Yes, AP7000 have two Ethernet MACs. And if I got this right you want to make a generic config about it, so then I guess it should open up for having more than one MAC.
OK, how about adding a CONFIG_MACB_ADVERTISE(id)-option, where id is the id of the MACB (passed to the macb_eth_initialize-function). This makes it possible to add this without touching anything but the macb-driver (i.e. without changing the macb_eth_initialize-prototype).
In the config-files, one could then have:
#define CONFIG_MACB_ADVERTISE(id) ( \ (id == 0) ? ( \ ADVERTISE_ALL | ADVERTISE_CSMA \ ) : ( \ ADVERTISE_CSMA | ADVERTISE_10HALF | \ ADVERTISE_10FULL \ ))
Or in the simple (and probably mose usual case (only one set of options advertised):
#define CONFIG_MACB_ADVERTISE(id) ( \
(ADVERTISE_CSMA | ADVERTISE_10HALF | ADVERTISE_10FULL)
This would require saving the id to the macb_device struct. If this is unacceptable, it could be changed to using the regs-offset instead of the id.
Any thoughts about this?
Best regards,
Olav Morken
PATCH v2 5/9 AVR32: macb - Disable 100mbps if clock is slow
Quoted 24 lines · Olav Morken
Quoted 3 lines
Yes, AP7000 have two Ethernet MACs. And if I got this right you want to make a generic config about it, so then I guess it should open up for having more than one MAC.
OK, how about adding a CONFIG_MACB_ADVERTISE(id)-option, where id is the id of the MACB (passed to the macb_eth_initialize-function). This makes it possible to add this without touching anything but the macb-driver (i.e. without changing the macb_eth_initialize-prototype).
In the config-files, one could then have:
#define CONFIG_MACB_ADVERTISE(id) ( \ (id == 0) ? ( \ ADVERTISE_ALL | ADVERTISE_CSMA \ ) : ( \ ADVERTISE_CSMA | ADVERTISE_10HALF | \ ADVERTISE_10FULL \ ))Or in the simple (and probably mose usual case (only one set of options advertised):
#defineCONFIG_MACB_ADVERTISE(id)( \
(ADVERTISE_CSMA|ADVERTISE_10HALF|ADVERTISE_10FULL)
Quoted 5 lines
This would require saving the id to the
macb_devicestruct. If this is unacceptable, it could be changed to using the regs-offset instead of the id.Any thoughts about this?
Sounds good to me. The board decides the id, so it makes sense to pass it back to the board code.
Haavard
PATCH v2 8/9 AVR32: CPU support for AT32UC3A0xxx CPUs
In reply to Wolfgang DenkQuoted 3 lines · Wolfgang Denk
Dear =?ISO-8859-1?Q?Gunnar_Rang=F8y?=,
In message <[email protected]> you wrote:
[....]
Quoted 18 lines
Quoted 8 lines
Quoted 3 lines
It would be nice if you used readable names instead of all these magic hardcoded constants.
Each bit in the masks represents a single pin on the microcontroller. The readable names would therefore become something like:
#defineEBI_PINS_PORT0(0x0003C000|0x1E000000)
#defineEBI_PINS_PORT1(0x00000010|0x00007C00| ...)Um, no. You didn't get it. You use magic numbers again.
That should be some
#define FOO_PIN_XXX 0x0003C000 #define FOO_PIN_YYY 0x1E000000 #define BAR_PIN_XXX 0x00000010 #DEFINE BAR_PIN_YYY 0x00007C00
Quoted 6 lines
in one place and
#define
EBI_PINS_PORT0(FOO_PIN_XXX|FOO_PIN_YYY) #defineEBI_PINS_PORT1((BAR_PIN_XXX|BAR_PIN_YYY)elsewhere, using readable names for the defines.
Sorry, but the bitwise or's in the code are a bit misleading. The only function of them is to make it more readable in the presence of the datasheet for the microcontroller. Each single bit in the bitmask is one pin on the microcontroller. In this case, it would become something like:
#define EBI_PIN_NCS0 (1<<14)
#define EBI_PIN_ADDR20 (1<<15)
#define EBI_PIN_ADDR21 (1<<16)
#define EBI_PIN_ADDR22 (1<<17)
(....)
#define EBI_PINS_PORT0 (EBI_PIN_NCS0 | EBI_PIN_ADDR20 | EBI_PIN_ADDR21 \
| EBI_PIN_ADDR22 | (....))Quoted 6 lines
Quoted 3 lines
I'm not certain that that would make it much more readable. The bitmasks are more or less based on reading the datasheet and counting which pins are used.
This is exactly what should NOT be necessary to read and understand the code.
[....]
But in this case, this is code which should never be changed without looking at the datasheet, and probably schematics for the board in question.
Quoted 8 lines
This is simply not acceptable.
Quoted 1 line
Maybe adding a comment before the code would be just as useful?
No, a comment can't fix this, especially as you just have to wait a few months to see the comment and the code getting out of sync.
Please fix the code.
How about something like this:
/*
* These bitmasks represents the pins used by the EBI on this board.
* Please refer to the datasheet for the UC3A-series microcontrollers
* for a description of the individual pins.
*/
#define EBI_PINS_PORT0 0x1E03C000
#define EBI_PINS_PORT1 0xE0037C10
(....)
And in the portmux-configuration-function:
portmux_select_peripheral(PORTMUX_PORT(0), EBI_PINS_PORT0,
PORTMUX_FUNC_C, 0);
(....)
When looking at this code I can see that we need a way for each individual board to change the bitmasks, since the same pin in some cases can be routed out to several places (depending on the board). Maybe a CONFIG_UC3A0xxx_EBI_PINS_PORT0 option or something like that could be added...
Quoted 11 lines
Quoted 8 lines
Quoted 2 lines
Instead of using offsets everywhere I would appreciate if the code could be changed to use C structs instead (like we do in PPC land).
It was done like this because it was done in the existing AVR32 code which we based our work on. Since we used parts of that code, it made most sense to use the same coding style.
We can change it if you think it is necessary.
Yes, please change it. And patches to fix the existing code that doe sthis are welcome, too.
OK, will do this. I do however think that we will leave the code for other chips and boards as it is, since changing it would require very much time, and we don't have any way to test the changes in any case.
Do you have any specific files we could use as a style-guide for such structs?
Quoted 7 lines
Quoted 4 lines
As most of the needed functionality is embedded in the microcontroller, there are very few external peripherals used by U-Boot. Apart from external memory, and oscillator, and level-shifters for the serial-port, there is only the ethernet PHY, and that one shouldn't need a reset.
Famous last words. What if exactrly the PHY is stuck and needs a reset?
The only reset we can do on the PHY is a software reset, by sending a reset command over the (R)MII bus, and I don't believe that the generic chip code is the place to do that. If it should be done, I believe it should be done by the macb-driver after the reset. This would allow it to recover even if the microcontroller wasn't reset by the reset-command, but for example by a watchdog timer.
Quoted 4 lines
Hmmm... "apart from external memory" ... does externam memory also include NOR flash? Eventually the NOR flash you are booting from? Assume the NOR flash is in query mode when you reset the board - how does it get reset, then?
External memory in this case would be SRAM or SDRAM.
Best regards,
Olav Morken
PATCH v2 6/9 AVR32: macb - Search for PHY id
In reply to Ben WarrenQuoted 1 line · Ben Warren
Added to net repo.
Great!
Does this mean that we shouldn't include this patch and the patch for volatile IP_t in the next version of the patch series?
Best regards,
Olav Morken
PATCH v2 8/9 AVR32: CPU support for AT32UC3A0xxx CPUs
In reply to Olav MorkenQuoted 35 lines · Olav Morken
Quoted 3 lines · Wolfgang Denk
Dear =?ISO-8859-1?Q?Gunnar_Rang=F8y?=,
In message <[email protected]> you wrote:
[....]
Quoted 18 lines
Quoted 8 lines
Quoted 3 lines
It would be nice if you used readable names instead of all these magic hardcoded constants.
Each bit in the masks represents a single pin on the microcontroller. The readable names would therefore become something like:
#defineEBI_PINS_PORT0(0x0003C000|0x1E000000)
#defineEBI_PINS_PORT1(0x00000010|0x00007C00| ...)Um, no. You didn't get it. You use magic numbers again.
That should be some
#define FOO_PIN_XXX 0x0003C000 #define FOO_PIN_YYY 0x1E000000 #define BAR_PIN_XXX 0x00000010 #DEFINE BAR_PIN_YYY 0x00007C00Quoted 6 lines
in one place and
#define
EBI_PINS_PORT0(FOO_PIN_XXX|FOO_PIN_YYY) #defineEBI_PINS_PORT1((BAR_PIN_XXX|BAR_PIN_YYY)elsewhere, using readable names for the defines.
Sorry, but the bitwise or's in the code are a bit misleading. The only function of them is to make it more readable in the presence of the datasheet for the microcontroller. Each single bit in the bitmask is one pin on the microcontroller. In this case, it would become something like:
Quoted 7 lines
#define EBI_PIN_NCS0 (1<<14)
#define EBI_PIN_ADDR20 (1<<15)
#define EBI_PIN_ADDR21 (1<<16)
#define EBI_PIN_ADDR22 (1<<17)
(....)
#define EBI_PINS_PORT0 (EBI_PIN_NCS0 | EBI_PIN_ADDR20 | EBI_PIN_ADDR21 \
| EBI_PIN_ADDR22 | (....))I don't think that would be much of an improvement. It would make the code much larger, and the pin definitions themselves would still need some magic number...and you'll need to know which GPIO port they belong to in order to actually use them.
Quoted 4 lines
Quoted 4 lines
Quoted 3 lines
I'm not certain that that would make it much more readable. The bitmasks are more or less based on reading the datasheet and counting which pins are used.
Quoted 8 lines
Quoted 2 lines
This is exactly what should NOT be necessary to read and understand the code.
[....]
But in this case, this is code which should never be changed without looking at the datasheet, and probably schematics for the board in question.
Exactly. At some point, you need code which encapsulates the definitions in the data sheet, and that's the whole purpose of these functions.
Quoted 21 lines
Quoted 8 lines
This is simply not acceptable.
Quoted 1 line
Maybe adding a comment before the code would be just as useful?
No, a comment can't fix this, especially as you just have to wait a few months to see the comment and the code getting out of sync.
Please fix the code.
How about something like this:
/*
* These bitmasks represents the pins used by the EBI on this board.
* Please refer to the datasheet for the UC3A-series microcontrollers
* for a description of the individual pins.
*/
#defineEBI_PINS_PORT00x1E03C000
#defineEBI_PINS_PORT10xE0037C10
(....)And in the portmux-configuration-function:
portmux_select_peripheral(PORTMUX_PORT(0),EBI_PINS_PORT0,
Quoted 8 lines
PORTMUX_FUNC_C, 0);(....)
When looking at this code I can see that we need a way for each individual board to change the bitmasks, since the same pin in some cases can be routed out to several places (depending on the board). Maybe a
CONFIG_UC3A0xxx_EBI_PINS_PORT0option or something like that could be added...
I think you need to actually use the parameters passed to that function in order to set up the correct address, data and chip select lines.
Now, the alternative mappings for some of the pins are tricky...but you could perhaps add a few more flags to cover those possibilities...
Once you do that, the whole function becomes a matter of shifting bits into position, and I very much doubt a bunch of defines can make that sort of code any clearer.
Think of this function as the executable equivalent of a pin definition.
Quoted 8 lines
Quoted 7 lines
Quoted 4 lines
As most of the needed functionality is embedded in the microcontroller, there are very few external peripherals used by U-Boot. Apart from external memory, and oscillator, and level-shifters for the serial-port, there is only the ethernet PHY, and that one shouldn't need a reset.
Famous last words. What if exactrly the PHY is stuck and needs a reset?
Quoted 6 lines
The only reset we can do on the PHY is a software reset, by sending a reset command over the (R)MII bus, and I don't believe that the generic chip code is the place to do that. If it should be done, I believe it should be done by the macb-driver after the reset. This would allow it to recover even if the microcontroller wasn't reset by the reset-command, but for example by a watchdog timer.
I suppose this might be a good idea. But I haven't heard of any problems with other boards because of this...and wasn't there a generic PHY layer in the works anyway?
Quoted 6 lines
Quoted 4 lines
Hmmm... "apart from external memory" ... does externam memory also include NOR flash? Eventually the NOR flash you are booting from? Assume the NOR flash is in query mode when you reset the board - how does it get reset, then?
External memory in this case would be SRAM or SDRAM.
On other chips, it also covers the NOR flash you're booting from. So I suppose we should look into this...maybe we need some sort of "notifier chain" thing to give other drivers a chance to reset their peripherals...
Haavard
PATCH v2 6/9 AVR32: macb - Search for PHY id
In reply to Olav MorkenQuoted 5 lines · Olav Morken
Quoted 2 lines · Ben Warren
Added to net repo.
Quoted 7 lines
Great!
Does this mean that we shouldn't include this patch and the patch for volatileIP_tin the next version of the patch series?Best regards,
Olav Morken
Correct. You've done a good job of making them orthogonal to the rest of your code, so the net repo is where they belong. I'll issue a pull request after doing a bit of testing.
regards,
Ben
Haavard Skinnemoen
AppliedPATCH v2 2/9 AVR32: Make cacheflush cpu-dependent
In reply to Gunnar RangoyQuoted 9 lines · Gunnar Rangoy
From: Olav Morken <[email protected]>
The AT32UC3A series of processors doesn't contain any cache, and issuing cache control instructions on those will cause an exception. This commit makes
cacheflush.harch-dependent in preparation for the AT32UC3A-support.Signed-off-by Gunnar Rangoy <[email protected]>Signed-off-by Paul Driveklepp <[email protected]>Signed-off-by Olav Morken <[email protected]>
Applied to evk1100-prep and merged into next. I'll send it upstream as soon as it's fine with Wolfgang.
Btw, I had to rebuild my next branch since it's become a bit stale. Since all the non-merge commit IDs are the same, I hope it won't cause any problems -- please let me know if you see any weird merge issues.
Haavard
Haavard Skinnemoen
AppliedPATCH v2 3/9 AVR32: Move addrspace.h to arch-directory, and move some functions from io.h to addrspace.h
In reply to Gunnar RangoyQuoted 11 lines · Gunnar Rangoy
From: Olav Morken <[email protected]>
The AVR32A architecture (which AT32UC3A-series is based on) has a different memory layout than the AVR32B-architecture. This patch moves
addrspace.hto an arch-dependent directory in preparation for AT32UC3A-support. It also moves some address-space manipulation functions fromio.htoaddrspace.h.Signed-off-by Gunnar Rangoy <[email protected]>Signed-off-by Paul Driveklepp <[email protected]>Signed-off-by Olav Morken <[email protected]>
Applied to evk1100-prep, thanks.
Haavard
Haavard Skinnemoen
AppliedPATCH v2 4/9 AVR32: Make GPIO implmentation cpu dependent
In reply to Gunnar RangoyQuoted 6 lines · Gunnar Rangoy
There are some differences in the implementation of GPIO in the at32uc chip compared to the ap700x series.
Signed-off-by Gunnar Rangoy <[email protected]>Signed-off-by Paul Driveklepp <[email protected]>Signed-off-by Olav Morken <[email protected]>
Applied to evk1100-prep, thanks.
Haavard
Haavard Skinnemoen
AppliedPATCH v2 7/9 AVR32: Must add NOPs after disabling interrupts for AT32UC3A0512ES
In reply to Gunnar RangoyQuoted 8 lines · Gunnar Rangoy
From: Olav Morken <[email protected]>
The AT32UC3A0512ES chip has a bug when disabling interrupts. As a workaround, two NOPs can be inserted.
Signed-off-by Gunnar Rangoy <[email protected]>Signed-off-by Paul Driveklepp <[email protected]>Signed-off-by Olav Morken <[email protected]>
Applied to evk1100-prep, thanks.
Haavard