From: u-boot-bounces@lists.denx.de [mailto:u-boot- Dear pshambhu,

In message 1380547665536-164381.post@n7.nabble.com you wrote:

...

As per previous posting i got to know that, there will be only one reset entry point, can't i have the another entry point in it.

You can talk to your chip vendor to provide you with some kind of logic to detect failed boot attempts and provide an alternative reset vecotr then. Guess your chances to gett that are extremely small, though.

You can, of course, throw hardware at it, and for example provide duplicate storage . memory devices for booting from, so you only have to swap chip select resp. address lines to select the alternative boot device.

...

• Why i can't have multiple entry points ? and what will be effect if i

introduced ?

You can't, because your processor only has one.

...

• Where the reset entry points are defined in the u-boot code ?

They are not defined in U-Boot, they are defined in the processor silicon or ROM code.

Disclaimer: of course things get even more complicated when you keep in mind that there is a multiverse of different hardware solutions, some of them with special, and others with truely exotic features.

you can tweak your hardware to split it, and many micro-controllers do it (especially for safety critical applications). Example: Suppose 'default' entry-point (or reset entry-point) = 0x0000_0000, Now OR your MSB bit with the fault-signal or boot-error-flag. reset-addr[31] = reset-addr_internal[31] | boot_error_flag; So in case when you primary boot fails, and boot_error_flags is set, next time when you boot, the address transforms into 0x1000_0000

This can also be done at board-level, where you can always re-route your default chip-select to some different memory by ORing it with boot_error_flag. But yes, you need a way to determine that your first boot failed, which is usually done by having a on-board watchdog, which timeout if system doesn't boot within given time.

with regards, pekon

Hi,

From: Wolfgang Denk [mailto:wd@denx.de] Dear "Gupta, Pekon",

In message 20980858CB6D3A4BAE95CA194937D5E73EA186A6@DBDE04.ent.ti.com

...

you can tweak your hardware to split it, and many micro-controllers do it (especially for safety critical applications). Example: Suppose 'default' entry-point (or reset entry-point) = 0x0000_0000, Now OR your MSB bit with the fault-signal or boot-error-flag.

Define "fault-signal" and "boot-error-flag". Thise may exist on your chip, but they don't on the overwhelming majority of systems.

No, these do not exist on my chip either :-). But you can have such functionality on-board using basic components like watchdog timers. Like some of Automobile safety systems have a separate on-board watchdog timer (apart from on-chip ones). And it is periodically patted via GPIO pins, by the software running on the system. But suppose a boot failed then this watchdog timer would expire and that timeout signal can be latched as boot_error_flag.

Case-1: This boot_error_flag can be used for re-routing chip-selects to other devices like (NAND), etc. Good part is such as this logic sits on board, it is independent of SoC. Bad part is it adds to your BOM cost.

Case-2: You can re-route boot_error_flag back to your SoC connecting it to NMI or external reset (most SoC would atleast have reset). And then a corrective action can be taken in your reset-entry handler, or exception handler. Jumps to exception-handlers and reset-handler will mostly execute because these branching are hard-coded in processor hardware.

Though I fully agree with you it's more difficult to do these things in actual than just describing the concept here.

with regards, pekon

Hi Wolfgang,

With respect to previous mail, i have one small doubt. I have two three u-boots,

u-boot_stub _ _|_ _ _ _ _ _ _ | | u-boot1 u-boot2

Can i have a small uboot_stub in the reset entry table, which will boot initially, and while booting it should select the default booting location (u-boot1).But if the default booting location fails to boot up then u-boot_stub should select bootloader u-boot2.

u-boot_stub will do CRC checksum on the u-boot1, if CRC checksum fails on that, then u-boot2 will be considered for bootup.

Since its the software, i think it can be done and If it is possible, can you please tell me which are the files need to be taken care for the file changes.

please guide me if i am wrong

Thanks & Regards Pradeep S

-- View this message in context: http://u-boot.10912.n7.nabble.com/Dual-boot-Images-in-Flash-tp164381p164443.... Sent from the U-Boot mailing list archive at Nabble.com.

Hi Wolfgang Denk,

Thanks for the feedback, for the customization of u-boot_stub, the main file start.S needs to be changed i guess, the u-boot_stub has to perform only CRC check. after performing it should goto u_boot1/u_boot2 address and then it should start booting.

How can i customize the u-boot_stub or generate the start.S file which perform only CRC calculation and switching address of u_boot1/u_boot2 , which performs booting.

Thanks & Regards Pradeep S

-- View this message in context: http://u-boot.10912.n7.nabble.com/Dual-boot-Images-in-Flash-tp164381p164588.... Sent from the U-Boot mailing list archive at Nabble.com.