Florin Iucha <florin at iucha.net> wrote ..
> Hello,
> 
> I have a unusual problem with putty connecting to several of my
> virtual servers.  It is so strange, I tried STFW with the symptoms but
> I can't find anything informative.
> 
> The client is a Windows 7 64-bit professional.  The servers are:
>    * Fedora 14 beta
>    * Centos 5.4 fully patched
>    * 3x Debian Squeeze/Testing (fully up-to-date)
> 
> When SSH-ing into the Centos and one of the Debian boxes I can use
> all the programs without any problem.  When SSH-ing into the Fedora
> and two of the Debian boxes, I can run programs like 'ls' and 'cat'
> that write small amounts of text to stdout, but if I start 'vim' or
> 'screen', I just get a blank screen.  Connecting again from a
> different machine and running an 'echo "t" > /proc/sysrq-trigger'
> I get this backtrace for the vim process:
> 
> [50583.985126] vim           S 0000000000000000     0  2613   2572 0x00000000
> [50583.985126]  ffffffff814611f0 0000000000000082 0000000000000000 ffffffff81191988
> [50583.985126]  0000000000000000 ffff88003f27b968 000000000000f9e0 ffff88003f27bfd8
> [50583.985126]  0000000000015780 0000000000015780 ffff88003f3fa350 ffff88003f3fa648
> [50583.985126] Call Trace:
> [50583.985126]  [<ffffffff81191988>] ? rb_insert_color+0x66/0xe2
> [50583.985126]  [<ffffffff810b809f>] ? zone_watermark_ok+0x20/0xb1
> [50583.985126]  [<ffffffff810649f7>] ? bit_waitqueue+0x10/0xa0
> [50583.985126]  [<ffffffff812fa693>] ? schedule_hrtimeout_range+0x4a/0x112
> [50583.985126]  [<ffffffff8105a50d>] ? del_timer_sync+0xc/0x16
> [50583.985126]  [<ffffffff810fb374>] ? poll_schedule_timeout+0x31/0x4f
> [50583.985126]  [<ffffffff810fbf13>] ? do_select+0x528/0x57a
> [50583.985126]  [<ffffffff810fc3df>] ? __pollwait+0x0/0xd6
> [50583.985126]  [<ffffffff810fc4b5>] ? pollwake+0x0/0x5b
> [50583.985126]  [<ffffffff810fc4b5>] ? pollwake+0x0/0x5b
> [50583.985126]  [<ffffffffa019fc8a>] ? __ext4_journal_stop+0x63/0x69 [ext4]
> [50583.985126]  [<ffffffffa018bdb5>] ? ext4_da_write_end+0x1fc/0x25a [ext4]
> [50583.985126]  [<ffffffff810b4a11>] ? generic_file_buffered_write+0x1f5/0x278
> [50583.985126]  [<ffffffff810b4e45>] ? __generic_file_aio_write+0x25f/0x293
> [50583.985126]  [<ffffffff8104a139>] ? try_to_wake_up+0x2a7/0x2b9
> [50583.985126]  [<ffffffff810fc0e9>] ? core_sys_select+0x184/0x21e
> [50583.985126]  [<ffffffff81064ad6>] ? autoremove_wake_function+0x0/0x2e
> [50583.985126]  [<ffffffff810cc64b>] ? handle_mm_fault+0x3b8/0x80f
> [50583.985126]  [<ffffffff8106bfbb>] ? ktime_get_ts+0x68/0xb2
> [50583.985126]  [<ffffffff810fc3b6>] ? sys_select+0x92/0xbb
> [50583.985126]  [<ffffffff81010b42>] ? system_call_fastpath+0x16/0x1b
> 
> But why would this happen with only two machines and not with the
> other two?  All the Debian servers are virtual machines hosted on
> the Fedora box.  They were installed from the same netinst iso and
> updated just yesterday.
> 
> If I SSH into the Centos box and from there I SSH into any other box
> (including the two troublesome Debian boxes) everything works fine as
> expected.
> 
> Any ideas?
> 
> Cheers,
> florin

To me, the situation looks like a server bug that produces a hash check error. If you look around in the configuration, there is an option that you can click that allows you to comprehend some ssh servers hash check errors and be able to access them even if the server has a hash-check error. I had the same problem with my phone, and when I asked someone to change my ssh server to comprehend the 2 alternate | characters in doing hash checks, along with the other alternated characters, that appears to have worked perfectly.