Recent versions broke compat with U-Boot "Jump to Script" feature

[Pierre Bourdon]

U-Boot supports SDP jumping to something that is not an IMX header. If that happens, it tries interpreting the data as a U-Boot script:

/* If imx header fails, try some U-Boot specific headers */
if (status) {
#ifdef CONFIG_SPL_BUILD
	/* In SPL, allow jumps to U-Boot images */
	...
#else
	/* In U-Boot, allow jumps to scripts */
	source(sdp_func->jmp_address, "script@1");
#endif
}

Unfortunately, recent versions of imx_usb_loader broke this by always trying to append a header and jumping to that header. I'm talking specifically about this in imx_sdp.c:

ld.dladdr = curr->load_addr;
ld.header_addr = ld.dladdr + ld.max_length;
ld.header_offset = curr->load_skip + ld.max_length;

...

ret = jump(dev, ld.header_addr);

Note: this new feature seems to be broken in my tests anyway -- the header was never transferred to the device because a bounds check later on caps the transfer size to the original file size.

Could you provide a way to restore the previous behaviour of transferring and jumping to headerless data?

Thanks!

[Troy Kisky]

On 4/23/2018 8:48 PM, Pierre Bourdon wrote: U-Boot supports SDP jumping to something that is not an IMX header. If that happens, it tries interpreting the data as a U-Boot script: |/* If imx header fails, try some U-Boot specific headers */ if (status) { #ifdef CONFIG_SPL_BUILD /* In SPL, allow jumps to U-Boot images */ ... #else /* In U-Boot, allow jumps to scripts */ source(sdp_func->jmp_address, ***@***.***"); #endif } | Unfortunately, recent versions of imx_usb_loader broke this by always trying to append a header and jumping to that header. I'm talking specifically about this in imx_sdp.c: |ld.dladdr = curr->load_addr; ld.header_addr = ld.dladdr + ld.max_length; ld.header_offset = curr->load_skip + ld.max_length; ... ret = jump(dev, ld.header_addr); | Note: this new feature seems to be broken in my tests anyway -- the header was never transferred to the device because a bounds check later on caps the transfer size to the original file size. Could you provide a way to restore the previous behaviour of transferring and jumping to headerless data? Thanks!

Thanks for the detailed report. I did not even know about the script support. I guess that is why someone broke the header support. I'll take a look at this after reviewing Stefan's changes. BR Troy

[Troy Kisky]

Hi Pierre Can you replace "jump" with "jump_direct" and see if it is working for you now ? Thanks Troy

…

On 4/23/2018 8:48 PM, Pierre Bourdon wrote: U-Boot supports SDP jumping to something that is not an IMX header. If that happens, it tries interpreting the data as a U-Boot script: |/* If imx header fails, try some U-Boot specific headers */ if (status) { #ifdef CONFIG_SPL_BUILD /* In SPL, allow jumps to U-Boot images */ ... #else /* In U-Boot, allow jumps to scripts */ source(sdp_func->jmp_address, ***@***.***"); #endif } | Unfortunately, recent versions of imx_usb_loader broke this by always trying to append a header and jumping to that header. I'm talking specifically about this in imx_sdp.c: |ld.dladdr = curr->load_addr; ld.header_addr = ld.dladdr + ld.max_length; ld.header_offset = curr->load_skip + ld.max_length; ... ret = jump(dev, ld.header_addr); | Note: this new feature seems to be broken in my tests anyway -- the header was never transferred to the device because a bounds check later on caps the transfer size to the original file size. Could you provide a way to restore the previous behaviour of transferring and jumping to headerless data? Thanks! — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#74>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABoNvyW8xIaIxvXQMD2P0Oz1oGYWNcosks5trqB7gaJpZM4Tg9ZY>.

[Pierre Bourdon]

Haven't tested it myself but 2 people reported it working on fail0verflow/shofel2#2

Thanks a lot for a super prompt response to this issue!
Best,