[aur-general] Sha256sum unexpected behaviour
Hello,
I got a weird issue with sha256sum behaviour. When creating the package with makepkg --geninteg I get a hash and when trying to build it with makepkg the hash fails. The interesting bit is --ptintsrcinfo comes up with a different hash.
Anyone got some idea why this might happen?
See below some debugging info
~/linux-gc$ makepkg --geninteg ==> Retrieving sources... -> Found linux-v5.4.10-arch1.tar.gz -> Found config -> Found 0001_bmq_v5.4-r1.patch -> Found 0002_enable_additional_cpu_optimizations-20190822.tar.gz ==> Generating checksums for source files... sha256sums=('bd51016d038277d2edd4e6d6def53b0ba24e0b238662d404a07a7d706ad2b5d3' 'aa77f7e3b611787f734c7e7d7503d3debf9af86ac999e8bccf533c5a854ec15b' '0b770209a72171dfd46401d67d13204a767eb1749ef522df7ea7292b83046537' '8c11086809864b5cef7d079f930bd40da8d0869c091965fa62e95de9a0fe13b5')
~/linux-gc$ makepkg --printsrcinfo pkgbase = linux-gc pkgdesc = Linux pkgver = 5.4.10 pkgrel = 1 url = https://cchalpha.blogspot.co.uk/ arch = x86_64 license = GPL2 makedepends = bc makedepends = kmod makedepends = libelf makedepends = xmlto makedepends = python-sphinx makedepends = python-sphinx_rtd_theme makedepends = graphviz makedepends = imagemagick makedepends = git options = !strip source = linux-v5.4.10-arch1.tar.gz::https://git.archlinux.org/linux.git/snapshot/linux-v5.4.10-arch1.tar.gz source = config source = 0001_bmq_v5.4-r1.patch::https://gitlab.com/alfredchen/bmq/raw/master/5.4/bmq_v5.4-r1.patch source = 0002_enable_additional_cpu_optimizations-20190822.tar.gz::https://github.com/graysky2/kernel_gcc_patch/archive/20190822.tar.gz validpgpkeys = ABAF11C65A2970B130ABE3C479BE3E4300411886 validpgpkeys = 647F28654894E3BD457199BE38DBBDC86092693E validpgpkeys = 8218F88849AAC522E94CF470A5E9288C4FA415FA sha256sums = 1fda6369ec7038c9ca1fe7aa6206977c5b8983c77e9d74e4ba458a4d042d1d31 sha256sums = aa77f7e3b611787f734c7e7d7503d3debf9af86ac999e8bccf533c5a854ec15b sha256sums = 0b770209a72171dfd46401d67d13204a767eb1749ef522df7ea7292b83046537 sha256sums = 8c11086809864b5cef7d079f930bd40da8d0869c091965fa62e95de9a0fe13b5
pkgname = linux-gc pkgdesc = The Linux kernel and modules with the BMQ CPU scheduler depends = coreutils depends = kmod depends = initramfs optdepends = crda: to set the correct wireless channels of your country optdepends = linux-firmware: firmware images needed for some devices provides = linux-gc=5.4.10
pkgname = linux-gc-headers pkgdesc = Headers and scripts for building modules for the Linux kernel depends = linux-gc provides = linux-gc-headers=5.4.10 provides = linux-headers=5.4.10
Thanks,
On Fri, Jan 10, 2020 at 08:27:11 +0000, Leonidas Spyropoulos via arch-general wrote:
Hello,
I got a weird issue with sha256sum behaviour. When creating the package with makepkg --geninteg I get a hash and when trying to build it with makepkg the hash fails. The interesting bit is --ptintsrcinfo comes up with a different hash.
Anyone got some idea why this might happen?
See below some debugging info
~/linux-gc$ makepkg --geninteg
Do you redirect the --geninteg output into the PKGBUILD? If not, the PKGBUILD is not updated, which would explain all the rest.
Alternatively, have a look at updpkgsums(8).
Best, Tinu
Anyone got some idea why this might happen?
makepkg --printsrcinfo does not compute anything, it just extracts the information from the PKGBUILD file. This is used to generate the .SRCINFO file.
makepkg --geninteg, on the other hand, downloads all the files in source array, computes the checksums and prints them. But it does not modify the PKGBUILD file.
So the different outputs are due to discrepancies between the hashes you computed locally and the ones in the PKGBUILD specified by the maintainer of that package in AUR.
Probably he forgot to add the PKGBUILD file to his last commit because he modified the .SRCINFO with the right hash.
Leave him a comment in AUR.
On 10 January 2020 08:27:11 GMT, Leonidas Spyropoulos via arch-general arch-general@archlinux.org wrote:
Hello,
I got a weird issue with sha256sum behaviour. When creating the package with makepkg --geninteg I get a hash and when trying to build it with makepkg the hash fails. The interesting bit is --ptintsrcinfo comes up with a different hash.
Anyone got some idea why this might happen?
See below some debugging info
~/linux-gc$ makepkg --geninteg ==> Retrieving sources... -> Found linux-v5.4.10-arch1.tar.gz -> Found config -> Found 0001_bmq_v5.4-r1.patch -> Found 0002_enable_additional_cpu_optimizations-20190822.tar.gz ==> Generating checksums for source files... sha256sums=('bd51016d038277d2edd4e6d6def53b0ba24e0b238662d404a07a7d706ad2b5d3' 'aa77f7e3b611787f734c7e7d7503d3debf9af86ac999e8bccf533c5a854ec15b' '0b770209a72171dfd46401d67d13204a767eb1749ef522df7ea7292b83046537' '8c11086809864b5cef7d079f930bd40da8d0869c091965fa62e95de9a0fe13b5')
~/linux-gc$ makepkg --printsrcinfo pkgbase = linux-gc pkgdesc = Linux pkgver = 5.4.10 pkgrel = 1 url = https://cchalpha.blogspot.co.uk/ arch = x86_64 license = GPL2 makedepends = bc makedepends = kmod makedepends = libelf makedepends = xmlto makedepends = python-sphinx makedepends = python-sphinx_rtd_theme makedepends = graphviz makedepends = imagemagick makedepends = git options = !strip source = linux-v5.4.10-arch1.tar.gz::https://git.archlinux.org/linux.git/snapshot/linux-v5.4.10-arch1.tar.gz source = config source = 0001_bmq_v5.4-r1.patch::https://gitlab.com/alfredchen/bmq/raw/master/5.4/bmq_v5.4-r1.patch source = 0002_enable_additional_cpu_optimizations-20190822.tar.gz::https://github.com/graysky2/kernel_gcc_patch/archive/20190822.tar.gz validpgpkeys = ABAF11C65A2970B130ABE3C479BE3E4300411886 validpgpkeys = 647F28654894E3BD457199BE38DBBDC86092693E validpgpkeys = 8218F88849AAC522E94CF470A5E9288C4FA415FA sha256sums = 1fda6369ec7038c9ca1fe7aa6206977c5b8983c77e9d74e4ba458a4d042d1d31 sha256sums = aa77f7e3b611787f734c7e7d7503d3debf9af86ac999e8bccf533c5a854ec15b sha256sums = 0b770209a72171dfd46401d67d13204a767eb1749ef522df7ea7292b83046537 sha256sums = 8c11086809864b5cef7d079f930bd40da8d0869c091965fa62e95de9a0fe13b5
pkgname = linux-gc pkgdesc = The Linux kernel and modules with the BMQ CPU scheduler depends = coreutils depends = kmod depends = initramfs optdepends = crda: to set the correct wireless channels of your country optdepends = linux-firmware: firmware images needed for some devices provides = linux-gc=5.4.10
pkgname = linux-gc-headers pkgdesc = Headers and scripts for building modules for the Linux kernel depends = linux-gc provides = linux-gc-headers=5.4.10 provides = linux-headers=5.4.10
Thanks,
makepkg --geninteg doesn't change the PKGBUILD, I think you're looking for 'updpkgsums'.
participants (4)
-
Iyán Méndez Veiga
-
Leonidas Spyropoulos
-
Tinu Weber
-
WorMzy Tykashi