Re: [PATCH v1] random: block in /dev/urandom

[Guenter Roeck <linux@roeck-us.net>]

On Fri, Apr 22, 2022 at 03:42:46PM +0200, Jason A. Donenfeld wrote:
> Hey Guenter,
> 
> On Tue, Mar 22, 2022 at 6:56 PM Guenter Roeck <linux@roeck-us.net> wrote:
> >
> > On 3/22/22 10:09, Jason A. Donenfeld wrote:
> > > Hey Guenter,
> > >
> > > On Tue, Mar 22, 2022 at 08:58:20AM -0700, Guenter Roeck wrote:
> > >> On Thu, Feb 17, 2022 at 05:28:48PM +0100, Jason A. Donenfeld wrote:
> > >>> This topic has come up countless times, and usually doesn't go anywhere.
> > >>> This time I thought I'd bring it up with a slightly narrower focus,
> > >>> updated for some developments over the last three years: we finally can
> > >>> make /dev/urandom always secure, in light of the fact that our RNG is
> > >>> now always seeded.
> > >>>
> > >>
> > >> [ ... ]
> > >>
> > >> This patch (or a later version of it) made it into mainline and causes a
> > >> large number of qemu boot test failures for various architectures (arm,
> > >> m68k, microblaze, sparc32, xtensa are the ones I observed). Common
> > >> denominator is that boot hangs at "Saving random seed:". A sample bisect
> > >> log is attached. Reverting this patch fixes the problem.
> > >
> > > As Linus said, it was worth a try, but I guess it just didn't work. For
> > > my own curiosity, though, do you have a link to those QEMU VMs you could
> > > share? I'd sort of like to poke around, and if we do ever reattempt this
> > > sometime down the road, it seems like understanding everything about why
> > > the previous time failed might be a good idea.
> > >
> >
> > Everything - including the various root file systems - is at
> > git@github.com:groeck/linux-build-test.git. Look into rootfs/ for the
> > various boot tests. I'll be happy to provide some qemu command lines
> > if needed.
> 
> I've been playing with a few things, and I'm wondering how close I am
> to making this problem go away. I just made this branch:
> https://git.kernel.org/pub/scm/linux/kernel/git/crng/random.git/log/?h=jd/for-guenter
> 
> Any interest in setting your tests on that and seeing if it still
> breaks? Or, perhaps better, do you have a single script that runs all

Looks like your code is already in -next; I see the same failures in
your tree and there.

openrisc generates a warning backtrace.

WARNING: CPU: 0 PID: 0 at drivers/char/random.c:1006 rand_initialize+0x148/0x174
Missing cycle counter and fallback timer; RNG entropy collection will consequently suffer.

parisc crashes.

[    0.000000] Kernel Fault: Code=15 (Data TLB miss fault) at addr 00000000
[    0.000000] CPU: 0 PID: 0 Comm: swapper Not tainted 5.18.0-rc3-32bit+ #1
[    0.000000]
[    0.000000]      YZrvWESTHLNXBCVMcbcbcbcbOGFRQPDI
[    0.000000] PSW: 00000000000001001011111100001110 Not tainted
[    0.000000] r00-03  0004bf0e 10f2c978 10773aa0 10e74300
[    0.000000] r04-07  00000004 10e74208 10e869d0 10e83978
[    0.000000] r08-11  f0023b90 f0023390 0004000e 10104f68
[    0.000000] r12-15  00000002 00000000 00000008 fffffff9
[    0.000000] r16-19  00000028 00080000 00000000 10dc6364
[    0.000000] r20-23  10dc6364 00000000 00000000 fefefeff
[    0.000000] r24-27  00000000 00000004 00000000 10dc6178
[    0.000000] r28-31  0073a08d 80000000 10e74340 00000000
[    0.000000] sr00-03  00000000 00000000 00000000 00000000
[    0.000000] sr04-07  00000000 00000000 00000000 00000000
[    0.000000]
[    0.000000] IASQ: 00000000 00000000 IAOQ: 1024d09c 1024d0a0
[    0.000000]  IIR: 0f401096    ISR: 00000000  IOR: 00000000
[    0.000000]  CPU:        0   CR30: 10e869d0 CR31: 00000000
[    0.000000]  ORIG_R28: 10e83ce8
[    0.000000]  IAOQ[0]: random_get_entropy_fallback+0x18/0x38
[    0.000000]  IAOQ[1]: random_get_entropy_fallback+0x1c/0x38
[    0.000000]  RP(r2): add_device_randomness+0x30/0xc8
[    0.000000] Backtrace:
[    0.000000]  [<10773aa0>] add_device_randomness+0x30/0xc8
[    0.000000]  [<10108734>] collect_boot_cpu_data+0x44/0x270
[    0.000000]  [<10104f28>] setup_arch+0x98/0xd4
[    0.000000]  [<10100a90>] start_kernel+0x8c/0x6d0

s390 crashes silently, no crash log.

Hope that helps,
Guenter