This is the default NIC emulated by QEMU.

I don't use.

We have EXT4 which can work with both ext2 and ext3. Besides that I
don't use any ext filesystems currently as root under QEMU is mounted
over 9P.

There is no leds to control from under QEMU.

There is no backlight to control under QEMU.

We have VIRTIO_GPU to be used under QEMU.

QEMU does not support joysticks and I do not need it at all.

Under QEMU we have network support via VIRTIO_NET.

There is no need to load firmware under qemu.

No need to load CPU microcode under QEMU.

We are running "QEMU Virtual CPU", not Intel.

There is no possibility a virtual CPU could get overheaten on something
similar.

+ enable CUSE (this is needed to test OSSPD).

CONFIG_KASAN_STACK=1 is selected automatically on GCC:

	git.kernel.org/linus/6baec880d7

Debian's cpuburn was removed from the distribution.

  qemu_debug += BPF
  qemu_debug: `make oldconfig` on 5.0.0-rc6 + fuse+next (v5.0-rc6-25-gfabf7e0262d0)
  qemu_debug += FUSE
  qemu_debug: `make oldconfig` on 5.0.0-rc1 + fuse-next (fuse-fixes-5.0-rc6-25-ge057f7fc89c2)
  qemu_debug: `make oldconfig` on 4.19.0-rc5
  mini64(3.16): tun: m->y; sit=y->m; increase log buffer size
  mini64(3.16): sky2: m -> y
  mini64(3.16): Align more networking options with lab.nexedi.com backend
  mini64(3.16): Try to align IPv6 options with lab.nexedi.com backend
  mini64: Start 3.16.7-ckt20 config
  mini64: CONFIG_IPV6_SUBTREES=y  (to run re6st)
  mini64: CFG80211_WEXT=y
  mini64: Try to enable BRCMSMAC
  mini64: +ext4 +ipv6 +namespaces (and in particular +userns)
  mini64: Make oldconfig for v3.17.1
  mini64: CONFIG_FHANDLE=y
  qemu_debug: Turn on all virtio
  qemu_debug: Kernel runs under qemu and logs to serial console
  qemu_debug: Start from mini64.conf and oldconfig it for v3.17-rc1-22-g480cadc
  mini64: make oldconfig for ~ 3.16.0-rc7
  ...

To be able to use e.g. bpftrace.

To debug fuse itself.

Inspired by gendiff from RPM. Defaults to .nxd0 suffix but suffix can be
set explicitly. See details in script header how to use & make local
changes so gendiff-nxd0 works.

Иначе без модулей не работает сеть.

То, что было =m, не включал (и проверил 95%) что модули не загружены.
То, что было =y - включил.

The same one as using on lab.nexedi.com backend.

Currently only oldconfig-extracted from latest mini64.conf.
Will adjust in following commits.

Spot in comments to recent LWN article.

- nice -19 for cpu
- ionice with idle scheduling class

From 2010. Needs fixing with recent Git. And better be upstreamed

http://git.vger.kernel.narkive.com/SoHylGJE/rfc-patch-git-amend-darcs-like-amend-record-for-git

To print only direct dependecies of an elf.

Например в ZODB есть ZODB.FileStorage.fsdump который в ZODB/setup.py
поставлен как

    entry_points = """
    [console_scripts]
    fsdump = ZODB.FileStorage.fsdump:main
    ...

т.е. после сборки оно сгенерирует отдельный файл со следующим содержимым

    #!/usr/bin/python
    # EASY-INSTALL-ENTRY-SCRIPT: 'ZODB3==3.9.7','console_scripts','fsdump'
    __requires__ = 'ZODB3==3.9.7'
    import sys
    from pkg_resources import load_entry_point

    if __name__ == '__main__':
        sys.exit(
            load_entry_point('ZODB3==3.9.7', 'console_scripts', 'fsdump')()
        )

т.е. программу можно будет запустить.

Но это после сборки. А in-tree программу не запустить.

Не уверен что добавление if __name__ == '__main__' примут в upstream
из-за entrypoint'а. Мне зачастую удобно работать in-tree, поэтому
добавляю простую утилиту для запуска таких программ.