Initramfs unpacking failed: junk in compressed archive

[cleesmith]

Following this article:
https://techblog.lankes.org/2015/05/01/My-Memo-to-build-a-custom-Linux-Kernel-for-Qemu/
then this one:
https://techblog.lankes.org/2015/07/05/Lightweight-Virtualization-on-OS-X/

I get a linux kernel 4.3.0 and initramfs which works using qemu:
qemu-system-x86_64 -smp 2 -kernel ./bzImage -initrd ./myinitrd.cpio -append "root=/dev/ram0 rootfstype=ramfs init=init console=ttyS0" -nographic

... but fails using xhyve:
../build/xhyve -m 1G -c 2 -s 0:0,hostbridge -s 31,lpc -l com1,stdio -f kexec,bzImage,myinitrd.cpio,"earlyprintk=serial root=/dev/ram0 rootfstype=ramfs init=init console=ttyS0"
...or:
../build/xhyve -m 1G -c 2 -s 0:0,hostbridge -s 31,lpc -l com1,stdio -f kexec,bzImage,myinitrd.cpio,"earlyprintk=serial console=ttyS0"

... with this message:

[    0.327131] Initramfs unpacking failed: junk in compressed archive
vm exit[0]
reason      VMX rip     0xffffffff81878ed0
inst_length 3
status      0
exit_reason 2
qualification   0x0000000000000000
inst_type       0
inst_error      0
./my_xhyverun.sh: line 27:  5078 Abort trap: 6

... same result when myinitrd.cpio is gzipped to myinitrd.cpio.gz, probably doesn't matter.

OS X Yosemite 10.10.5
git clone https://github.com/mist64/xhyve.git
The included tiny core linux in the test folder works fine.
Any ideas as to why it's junk to one but not the other?
Did I miss something?
Thanks for this project.

[cleesmith]

It appears that 4.x kernels from https://www.kernel.org/ do not work with xhyve, odd, but pre-4 kernels work just fine. Thanks.

[luke-clifton]

I am seeing this issue, why was this closed? Is there a workaround?

[cleesmith]

There's no work around that I know of, and I never received a response to this issue so I closed it and moved on.

[xez]

Did you compile the kernel with CONFIG_KEXEC=y?

[peterwang]

I came across the same issue while installing NixOS. I solved it by adding an offset of 16777216 (16M) to ramdisk_start (in file src/firmware/kexec.c). btw: I got the 16M value by trial and error. Here is the patch for file kexec.c:

diff --git a/src/firmware/kexec.c b/src/firmware/kexec.c
index 61aeebb..b190289 100644
--- a/src/firmware/kexec.c
+++ b/src/firmware/kexec.c
@@ -186,6 +186,12 @@ kexec_load_ramdisk(char *path) {

        ramdisk_start = ALIGNUP((kernel.base + kernel.size), 0x1000ull);

+        /* temp hack */
+        char* tmpstr = getenv("RAMDISK_START");
+        if (tmpstr != NULL) {
+            ramdisk_start += ((uint32_t) atoi(tmpstr));
+        }
+
        if ((ramdisk_start + sz) > memory.size) {
                /* not enough memory */
                fclose(f);

[zchee]

@peterwang This problem has been fixed on docker/hyperkit
moby/hyperkit#66
https://github.com/docker/hyperkit/pull/66/files

[c4milo]

@zchee can it be back ported to xhyve?

[zchee]

@c4milo sure. I'll send a pull request.

[zchee]

@c4milo done :)

[c4milo]

Nice, thanks!!
On Fri, Oct 21, 2016 at 1:17 PM Koichi Shiraishi notifications@github.com
wrote:

@c4milo https://github.com/c4milo done :)

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#74 (comment), or mute
the thread
https://github.com/notifications/unsubscribe-auth/AACi0TRZykh3Q-mRqQm4dRn9kUOorKjNks5q2POcgaJpZM4GeuXj
.