On 30 October 2017 at 15:33, Gregory CLEMENT <gregory.clem...@free-electrons.com> wrote: > Hi Ard, > > On lun., oct. 30 2017, Ard Biesheuvel <ard.biesheu...@linaro.org> wrote: > >> On 30 October 2017 at 15:09, Gregory CLEMENT >> <gregory.clem...@free-electrons.com> wrote: >>> Hi Ard, >>> >>> On lun., oct. 30 2017, Ard Biesheuvel <ard.biesheu...@linaro.org> wrote: >>> >>>> On 30 October 2017 at 15:05, Gregory CLEMENT >>>> <gregory.clem...@free-electrons.com> wrote: >>>>> Hi Ard, >>>>> >>>>> On lun., oct. 30 2017, Ard Biesheuvel <ard.biesheu...@linaro.org> wrote: >>>>> >>>> ... >>>>>> >>>>>> Could you please share the output of 'readelf -S' for those vmlinux >>>>>> decompressor images? >>>>> >>>>> Here it is: >>>>> >>>>> In the meantime I also used an arm-linux-gnueabihf- in case it could be >>>>> related to the toolchain, I had te same issue and here it is the readelf >>>>> output: >>>>> arm-linux-gnueabi-readelf -S ../build/vmlinux >>>> >>>> Actually, I meant the 'vmlinux' file in arch/arm/boot/compressed, not >>>> the main one. >>> >>> Here it is then: >>> >>> arm-linux-gnueabi-readelf -S ../build/arch/arm/boot/compressed/vmlinux >>> There are 22 section headers, starting at offset 0x4b3e14: >>> >>> Section Headers: >>> [Nr] Name Type Addr Off Size ES Flg Lk >>> Inf Al >>> [ 0] NULL 00000000 000000 000000 00 0 >>> 0 0 >>> [ 1] .text PROGBITS 00000000 010000 005ef8 00 AX 0 >>> 0 32 >>> [ 2] .rodata PROGBITS 00005ef8 015ef8 000d4c 00 A 0 >>> 0 4 >>> [ 3] .piggydata PROGBITS 00006c44 016c44 48a33a 00 A 0 >>> 0 1 >>> [ 4] .got.plt PROGBITS 00490f80 4a0f80 00000c 04 WA 0 >>> 0 4 >>> [ 5] .got PROGBITS 00490f8c 4a0f8c 000028 00 WA 0 >>> 0 4 >>> [ 6] .pad PROGBITS 00490fb4 4a0fb4 000004 00 WA 0 >>> 0 1 >>> [ 7] .bss NOBITS 00490fb8 4a0fb8 00001c 00 WA 0 >>> 0 4 >>> [ 8] .stack NOBITS 00490fd8 4a0fb8 001000 00 WA 0 >>> 0 1 >>> [ 9] .comment PROGBITS 00000000 4a0fb8 00001c 01 MS 0 >>> 0 1 >>> [10] .ARM.attributes ARM_ATTRIBUTES 00000000 4a0fd4 00002d 00 0 >>> 0 1 >>> [11] .debug_line PROGBITS 00000000 4a1001 00281b 00 0 >>> 0 1 >>> [12] .debug_info PROGBITS 00000000 4a381c 0066cb 00 0 >>> 0 1 >>> [13] .debug_abbrev PROGBITS 00000000 4a9ee7 0013ea 00 0 >>> 0 1 >>> [14] .debug_aranges PROGBITS 00000000 4ab2d8 0001a8 00 0 >>> 0 8 >>> [15] .debug_str PROGBITS 00000000 4ab480 0019b4 01 MS 0 >>> 0 1 >>> [16] .debug_ranges PROGBITS 00000000 4ace38 000640 00 0 >>> 0 8 >>> [17] .debug_frame PROGBITS 00000000 4ad478 001010 00 0 >>> 0 4 >>> [18] .debug_loc PROGBITS 00000000 4ae488 003643 00 0 >>> 0 1 >>> [19] .symtab SYMTAB 00000000 4b1acc 0015b0 10 20 >>> 225 4 >>> [20] .strtab STRTAB 00000000 4b307c 000cc5 00 0 >>> 0 1 >>> [21] .shstrtab STRTAB 00000000 4b3d41 0000d2 00 0 >>> 0 1 >>> Key to Flags: >>> W (write), A (alloc), X (execute), M (merge), S (strings), I (info), >>> L (link order), O (extra OS processing required), G (group), T (TLS), >>> C (compressed), x (unknown), o (OS specific), E (exclude), >>> y (purecode), p (processor specific) >>> >> >> And this is from the build the build that generated the linker assert? > > Humm no, actually it was with the wrong branch. If I have the patch > "ARM: verify size of zImage" then arch/arm/boot/compressed/vmlinux is > not generated. > > But if I remove this patch then I can generate this file and so: > arm-linux-gnueabi-readelf -S ../build/arch/arm/boot/compressed/vmlinux > There are 22 section headers, starting at offset 0x4b402c: > > Section Headers: > [Nr] Name Type Addr Off Size ES Flg Lk Inf > Al > [ 0] NULL 00000000 000000 000000 00 0 0 > 0 > [ 1] .text PROGBITS 00000000 010000 005ef8 00 AX 0 0 > 32 > [ 2] .rodata PROGBITS 00005ef8 015ef8 000d4c 00 A 0 0 > 4 > [ 3] .piggydata PROGBITS 00006c44 016c44 48a552 00 A 0 0 > 1 > [ 4] .got.plt PROGBITS 00491198 4a1198 00000c 04 WA 0 0 > 4 > [ 5] .got PROGBITS 004911a4 4a11a4 000028 00 WA 0 0 > 4 > [ 6] .pad PROGBITS 004911cc 4a11cc 000004 00 WA 0 0 > 1 > [ 7] .bss NOBITS 004911d0 4a11d0 00001c 00 WA 0 0 > 4 > [ 8] .stack NOBITS 004911f0 4a11d0 001000 00 WA 0 0 > 1 > [ 9] .comment PROGBITS 00000000 4a11d0 00001c 01 MS 0 0 > 1 > [10] .ARM.attributes ARM_ATTRIBUTES 00000000 4a11ec 00002d 00 0 0 > 1 > [11] .debug_line PROGBITS 00000000 4a1219 00281b 00 0 0 > 1 > [12] .debug_info PROGBITS 00000000 4a3a34 0066cb 00 0 0 > 1 > [13] .debug_abbrev PROGBITS 00000000 4aa0ff 0013ea 00 0 0 > 1 > [14] .debug_aranges PROGBITS 00000000 4ab4f0 0001a8 00 0 0 > 8 > [15] .debug_str PROGBITS 00000000 4ab698 0019b4 01 MS 0 0 > 1 > [16] .debug_ranges PROGBITS 00000000 4ad050 000640 00 0 0 > 8 > [17] .debug_frame PROGBITS 00000000 4ad690 001010 00 0 0 > 4 > [18] .debug_loc PROGBITS 00000000 4ae6a0 003643 00 0 0 > 1 > [19] .symtab SYMTAB 00000000 4b1ce4 0015b0 10 20 225 > 4 > [20] .strtab STRTAB 00000000 4b3294 000cc5 00 0 0 > 1 > [21] .shstrtab STRTAB 00000000 4b3f59 0000d2 00 0 0 > 1 > >> >> Could you check the value of _edata in the output of 'nm vmlinux' >> please? > With this kernel: > > arm-linux-gnueabi-nm ../build/arch/arm/boot/compressed/vmlinux | grep _edata > 004911d0 D _edata >
Well, that is disappointing. This means the ASSERT() does not work reliably, and we're back to using a bunch of shell scripts to check whether _edata appears at the end of the image. Which version of binutils are you using?
