Posted on

FreeBSD 10.3 hangs on upgrade – beware!

There seems to be a bit of a problem with upgrades to FreeBSD 10.3-RELEASE. Basically, shutdown -r is hanging, requiring you to manually reset the machine (turn it off and on again). This is annoying unless the machine in question happens to be at a data centre on a different continent, in which “annoying” really doesn’t cut it.

This was a known issue with 10.3-STABLE., but it appears to have made it in to -RELEASE too.

I suggest not using freebsd-update. Basically if you follow the official instructions you may need someone on hand to reboot the old fashioned way.

Posted on

OKI laser duplex unit doesn’t work

OKI C5650 or C5750

This was a weird one. For some reason my OKI c5750 printer (similar to c5650) started ignoring its duplex unit. Stuff was printing on two sheets, single sided. I checked the Windows drivers, and that Duplex long-edge was enabled on them and on the printer control panel. But nothing was doing. The two single-sided sheets came out instead of double-sided, and what’s more it seemed slower than usual.

I initially thought it was a Windows fault introduced by my recent troubles. Had I restored the printer driver to a weird state? But after half an hour of fishing around I finally found the problem on the web interface. It’s subtle.

You can select the weight and type of paper for various trays. It turned out that the tray I was using was set to “thick” and “glossy”. I reset this to “normal weight” and “normal”, and everything started working again. I assume that the OKI internal software won’t send think paper (i.e. light card) into the duplexer, but it doesn’t tell you this. The Windows restore must have set the manual feed tray to light card. This would also explain the slower roller feed.

I hope this helps if you’re also having trouble.

Posted on

Great Northern Trains First Class grott-hole

GreatNortherFirstClassTip
Great Northern Trains “First Class”. What you can’t see is the dirt, sticky carpets and tables.

I’m currently on the 11:52 Great (sic) Northern Trains service from Kings Cross to Cambridge. I’m in First Class. The compartment is a tip – strewn with litter and the tables are filthy. It has the kind of carpet that you stick to in places. The ticket collector has just been and gone, and I pointed out the state of the place. She shrugged. I enquired if anyone ever cleaned it. “Only if there’s time.”, she said.

I pointed out that this really wasn’t good enough, especially for a first class compartment. It wasn’t Northern Train’s fault, came the reply. Apparently “It’s because of the customers, we put up notices but they still drop litter.”

In a normal compartment I’d have some sympathy with this; in First Class it’s not really the right attitude. I’ve seen at least three Northern Trains staff crewing the train; one of them could clear it. Why else am I paying three times the price for the ticket? For a table I don’t want to use without using anti-bacterial cleaner first (which I forgot to pack silly me).

Update:

I brought the state of the compartment to the attention couple of Northern Trains customer service representative ticket barrier. Apparently the turn-around at Kings Cross is too short to do anything about cleaning the whole train. I pointed out that I was talking about a dozen seats and two tables in a tiny First Class compartment and they fully understood my point, agreed it was indeed unacceptable, and apologised for the ticket inspector. So Norther Trains – are you listening?

Posted on

Microsoft Windows Backup Dramas

brokengreenglassAlways take a backup. Everyone knows that. There are plenty of articles on the web singing the virtues of the the new (Windows 7, 8 and 10) backup utility, and how easy it is to use. But none on how to actually restore your data when things go wrong. ‘sfunny that.

A week ago, when my Windows PC decided to scramble its hard disk I felt smug for having followed my long-standing advice. I’d taken an Image Backup of the entire system disk with the applications installed, data files were shadowed on a server (FreeBSD of course) and I had last month’s “User files” Windows backup to bring the system up-to-date (the odd Windows update notwithstanding). And even better, the system was still in a condition to boot albeit with a few warnings about corrupted DLLs, none of which were important.

Ha!

The first stage in any situation like this is to make an exact copy of the trashed drive in case things go horribly wrong. This involves mounting the disk on a working system and copying the drive as data. If you;re reading this because you are having problems, but don’t have the ability to image a drive, please find someone who can do this for you. The remainder of this diatribe assumes that you either know what you’re about, or at least have a safe copy of the disk you’re about to try and resurrect.

Only when the drive is imaged and safe is it okay to to try to boot the machine and run CHKDSK.  No hardware faults were found in this case, but it fixed numerous errors in the directories and it was pretty obvious that nothing on the disk could be trusted. Windows had obviously gone ape and trashed the disk itself. so it was necessary to restore from a Backup Image to ensure the system was clean. And this is where my fun really started.

You have to reboot the PC into recovery mode (Windows RE). This is achieved by holding down a function key on boot  and selecting the “Recovery” option from possible boot disks (details vary by manufacturer), or if you’re running Windows already you can find it under Advanced Options on System Restore – it will reboot to Windows RE mode for you. Then all you need to is select the image you want to restore, which in my case was on the network, so I entered the server details and login as requested. No dice.

Re image Your Computer

An internal error occurred. The following information
might help to resolve the error:

The network location cannot be reached. For information
about network troubleshooting, see Windows Help (0x0800704CF)

The restore utility must be wrong, and a quick Google search threw up some possible explanations. In my case it was nothing to do with the file permissions; it was on a FreeBSD cluster and I’m very confident I have control over the file permissions on that.

So what next? A Windows Image backup (since Windows 7) creates a .vhd file (Virtual Hard Disk). This is Microsoft’s half-arsed equivalent to being able to mount a file as an device and then using an FS on it. You can mount (Attach) a .vhd file from the “Mange Computer” console, using menu that pops up when you right-click on the “Disk Management” part of the tree on the left, which is not obvious. (Even less obvious: to detach it you need to right-click on the left-hand area representing the drive in the partition map on the right).

So, if Windows RE couldn’t read the image across the network I decided to use Windows itself to copy the files back manually by mounting the .vhd from the backup location. This isn’t that simple; you can’t copy the OS back while it’s running so you need to boot from somewhere else before you finish the job, but its better than a poke in the eye with a sharp stick.

Again, no dice.

“This image was created by a different version of Windows”

What? No way! It was created by this version of Windows; on this PC, no less.

This may have been because, for some bonkers reason, you can’t mount a backup .vhd as read-only, at least the first time its mounted on anything. It doesn’t tell you this, it just says that its an incompatible version. You must un-check the Read-Only box; and if you’re like me, you’ll make a backup of the backup before doing this. But it still didn’t work.

I tried again with the backup .vhd I’d made, largely because it had a much shorter name, and bingo! It mounted and all my files were visible. What’s going on? It wasn’t a permissions issue – of that I’m certain. The only explanation I’ve have is that the path name was too long before, with the Microsoft names being a UUID (GUID). Try shortening the name/path if you can’t mount a .vhd.

Armed with this knowledge that the .vhd really was good, I fiddled the file names and associated links and tried Windows RE image restore again, but with the same 0x800704CF error code as a result. A bit more digging, and it turns out this means that basically the system isn’t talking to the network card, probably because there is no driver. Microsoft’s solution is to install the driver with the disk that came with the card. But this Lenovo PC came with pre-installed Windows and NIC – Microsoft doesn’t supply Windows OEM on a disk any more, never mind a driver CD. One might assume, given that Windows RE happily asked me for network credentials as I supplied it with the network address of the image, that it had the network driver pre-installed. But that would be to sensible.

Rather than messing about looking for a suitable NIC driver and burning it to a CD, I just copied the “WindowsImageBackup” folder to the root of a USB-connected and disk and it restored just fine. Actually, it only worked after I moved the USB drive from its normally USB 3 slot to a USB 2 slot, because presumably Windows RE doesn’t have a driver for USB 3 hardware either. A pattern was emerging.

The image was restored; all my software was back. The only snag is that there were more than 200 Microsoft updates; taking another four hours to install. But at least I knot it’s now a clean install, and as a reward for my hard work I was able to put put of installing the Windows 10 “upgrade” nagware (KB2952664 and KB2976978 if you’re interested – it doesn’t identify itself as such).

So what have I learned from all of this? “Don’t Trust Microsoft”? Well I haven’t trusted a Microsoft backup solution since MS-DOS 2.1, and with good reason. But from now on I’ll always take the trouble to use dd image for imaging drives in future, even if that means taking them out of the PC to do it.

Posted on

FreeBSD Device Driver Memory Allocation

Yesterday someone asked me how to allocate memory in a FreeBSD device driver. Although not quite as simple as a user-space malloc(), it’s relatively simple – but could I remember the name/parameter order? Not confidently, so I suggested RTFM.

A quick look at the manual doesn’t actually cover it very well. Basically there are special versions of malloc()/free() and they’re have exactly the same names, except the parameters are different. For example, malloc() has two extra parameters; one is the memory type (used for kernal statistics purposes), and one is a flags field, with options whether you’re prepared to wait, or is this a critical situation and using the reserve pool is okay.

For details, see “man 9 malloc”. The ‘9’ is important, as otherwise you’ll get the user-land version in libc. (Incidentally, a read through the libc code should put you off algorithms making wanton use dynamic memory allocation if you weren’t already).

Now what the FreeBSD documentation doesn’t tell you (and something for my to-do list) is how to actually make use of this in a device driver. I had to go back to code I’d written ten years ago to remind me, as I’m just as guilty of copying and tweaking my standard code many times over without really remembering what it does.

But before you go worrying about allocating dynamic memory in a device driver, consider that there’s no reason why you can’t just use static memory – just allocate in BSS in the normal way. Okay, this won’t suit every eventuality but on on most of my simple drivers, which have been to mess with custom hardware for a single process, it’s not actually a problem.

Okay, so you still want to use dynamic? Well to get the kernel versions instead of the the libc ones you need to include instead. As I mentioned above, for some reason using the same names must have seemed like a good idea at the time, but the parameters are different.

The other thing you should be aware of is when about allocating kernel memory you are talking about non-paged. Don’t go crazy.

There is also a memory allocation tracker and statistics dumper available in the libc version (see /etc/malloc.conf), which will help you out if you’ve messed up memory allocation. Don’t expect any such help with the kernel. However, if you compile the kernel with the INVARIANTS option set it will scrub freed memory with 0xdeadc0de, which is handy if you find yourself using unallocated or free kernel RAM. Actually, this is a pretty good idea if you’re writing KLDs anyway, as it stops and does a core dump at the first sign you’ve screwed up any kernel structures.

The documentation in “man 9 malloc” should be enough to cope with the extra parameters; basically the malloc_type. Note that the first parameter to the MALLOC_DEFINE macro is actually a name you make up! By convention it’s in the form M_XXXXX, in upper case.

Also note that when you’re freeing memory it’s not normally zeroed. Therefore someone else using kernel memory might be able to allocate it and read what your driver wrote. Okay, bug deal – if the bad guys are installing kernel modules it’s game anyway. But… consider the bad guys cause a kernel panic and get a core dump.

 

Posted on

Flash Crash (Adobe version)

AGet the dobe Flash (the browser plug-in) is notorious as a security risk, and the current batch of known exploits does nothing to improve it’s reputation. Sorry Adobe.
CVE-2016-1010 is the latest biggie, as it allows remote code execution on all but the very latest plug-in. There’s also CVE-2015-8651, CVE-2015-7645, CVE-2016-0963 and CVE-2016-0993 to worry about.

You should, of course, make sure that you have the latest plugging installed on your browser. Unfortunately the version numbering system varies by platform so I can’t easily tell you which you need.

When looking at multifarious Adobe Flash vulnerabilities in the NIST database I’m always amused to note that it appears to be written in Coldfusion. For the last ten years that’s been Adobe Coldfusion. Oh my!

 

Posted on

BBC micro:bit finally launched

At verybbcmicrobit_s long last, the BBC micro:bit has been released. This is the educational embedded computer designed to inspire  kids to learn about real programming. A small board with a CPU, Bluetooth, two switches and some LEDs it’s ideal for… Well what? Obvious comparisons will be made with the established but overcomplicated Raspberry Pi.

The plan is to send these out to year 7 students over the Easter holiday. I’m involved in computer science education, but I can’t even buy one (although I can use the simulator). Quite how these will be received when they turn up during Summer term remains to be seen, but I suspect eBay will feature in getting them to those who are interested in this kind of thing.

Unfortunately, from it’s inception in 2012, those of us who have been watching events unfold have a one-word verdict in common: Fiasco.

I’ll let you know more if I actually get to see one.

Posted on

FreeBSD sysarch kernel panic vulnerability

A bug has been found and fixed in the FreeBSD kernel that would allow someone with malicious intent to crash a running system. It’d be difficult to achieve unless the attacker had console access. However it’s been patched for all supported systems. See here for all the details (which I won’t repeat).

The problem was found by Core Security, and they have provided an excellent write-up here.

But if you want it in plain English:

The sysarch() system call is used to get/set processor-specific stuff. You’re not supposed to call it directly; you’re supposed to call a processor-specific library if you want to do things like that, but you still can call it if you want to. On processors that support memory segments, such as i386,  there is a Local Descriptor Table (LDT) to manage them if you want to mess with specific stuff like that. However, for security reasons, you can only modify the LDT using the sysarch() call, which checks what you’re trying to do and prevents applications from doing anything crazy.

Unfortunately the AMD64 implementation of the code gets the checking wrong. If you use a signed integer it’s always going to be less than another unsigned value, and when it compares the two parameters to make sure that one is less than the other it passes when it shouldn’t, and the rogue parameter causes it to go funky-deux and overwrite a shed load of stuff.

This is in all in:

/sys/amd64/amd64/sys_machdep.c

in the function:

int amd64_set_ldt(td, uap, descs)

The FreeBSD advisory contains a patch for all “supported” versions; but what if you’re using an older one? Using the information from Core it’s easy enough to patch. But what else is affected?

To save you the trouble, I’ve looked back at earlier versions. The problem code definitely exists in the AMD64 versions for 8.x, but isn’t present in any 7.x, as far as I can tell. The system call simply doesn’t exist. On i386 versions, I can’t see any obvious problem with the code.

How worried should we be? If someone breaks in to a system with shell access, they will be able to crash it. However, I think it’s very unlikely that any service is written in such a way that malicious data could cause the necessary parameters to be sent to sysarch() call. In fact, on checking the ports collection, it’s not exactly used all over the place. You’re highly unlikely to be running any application that even makes the call.

Posted on

Android Stagefright bug gets serious

AndroidLogoSThere’s a bug in all by the most recent versions of the Android operating system that can theoretically allow attackers to take over the device simply by viewing a web page or downloading a media file. It’s actually in the Stagefright library, and was the talk of Black Hat last August. Then it was considered hard to exploit, but security researcher Hanan Be’er at  North-Bit in Israel has now published a paper proving it’s very dangerous.

Stagefright is the name of the media processing library found in all versions of Android you’re likely to find. It opens and reads any media downloaded to the device. With a specially crafted file you can cause it to crash when it does this; you don’t have to even play the file. However, it has been difficult to make use of this fact to “break out” and do anything more nasty.

Since Android 5.0, a system called Address Space Layout Randomisation (ASLR) has been in use. Basically the memory space is shuffled randomly so malicious code doesn’t know where anything else is, making attacks more difficult. This made exploiting Stagefright’s flaws a lot harder. The fact that the problem exists on Android 2.2 to 4.x, which doesn’t do ASLR, has been the subject of much complacency. Google has released fixes for the bug, known as CVE-2015-3864, but by no means have all the Android devices been updated. I guess that the vast majority have not, including the recent ones using Android 5.x. The infrastructure for updating Android simply doesn’t exist. Apple’s devices are very exploitable, but at least they have a mechanism for updating them.

So how does the North-Bit exploit work? It’s actually very straightforward. First you deliver a dodgy video file to the device; putting it on a web site is the obvious, easy method. This will cause Stagefright to crash and restart in a known state. When it does this, some JavaScript running on the same page slurps various parameters on the system, such as the current location of libc, and sends it back to the attacker. A new video file is then created and sent using this information, and it’s game over – possibly after a few tries, but North-Bit says the exploit is reliable.

How worried should we be? I’d say we should be very worried. Unless your device manufacturer and/or mobile network rolls out the patch, I can’t see any mitigation.

Posted on

Apple is too cool for the CIA to touch

Tim Cook 2009 cropped
Tim Cook – time he was sent to jail?
You can’t have missed the furore over Apple’s refusal to help the CIA get the data from a terrorist murderers iPhone. On the one side the CIA says that we need the data to protect the public, a line with the judiciary of the USA agrees with, and Apple should do everything possible to get it for them. On the other side there’s Apple’s PR engine trying (successfully) to spin the story and avoid complying with the court order.

In the mean time the Brazilians haven’t shown such deference to a cultural icon when it comes to Facebook owned WhatsApp refusing to hand over data concerning a major drugs trafficker, even after several court orders. The Brazilian authorities have arrested Diego Dzodan, Facebook’s hancho in Latin America, and thrown him in jail until such time as the company obeys the law.

Perhaps he Americans could try that with Tim Cook – you break the law, you go to jail.

Meanwhile, Apple might seem to be setting itself up as the criminals friend over this. In the land of the free where profit is king, I guess their money is as good as anyone else’s so perhaps we should be too judgemental. But in an outrageous spin, Apple has told the world that if they comply with the court order then all Apple handsets will have a backdoor and no longer be secure. This is disingenuous. The situation is this:

Apple encrypts the data stored on the phone. You have to enter a password to unlock it. If you enter ten wrong passwords it will wipe the data from the phone. The CIA has asked Apple to modify this handset to disable the data wiping feature, so the CIA can then just keep throwing passwords at it until it unlocks. Clearly, this is going to have no physical effect on any other handset anywhere else in the world. So what’s Apple’s problem?

If Apple helped the CIA break in to the handset, Apple can no longer claim that its handsets are invulnerable. Terrorists, fraudsters and anyone up to something will know that the authorities can get at Apple data even more easily than if it was stored on iCloud. Note well: the fact that Apple hasn’t produced the mod needed to do this (publicly), doesn’t mean that its not possible right now; and it may even be happening. But Apple wants to maintain the illusion that it can’t.

Put another way, it’s easy enough to bypass the locks on a front door. You just need a large enough sledge hammer. Doubt this? Look at the footage of a police raid taking place – a few burly coppers with a battering ram and it’s open in seconds. Apple is selling locks and trying to pretend there’s no such thing as a sledgehammer.

So why, might one ask, don’t the US authorities stop messing around and get the court order enforced? Are they really scared of Apple?

What’s really worrying about this situation is that “civil liberties campaigners” and some corporate America is rushing to put out statements in Apple’s defence. In other words, big business reckons it’s above the law made by the people using a democratically elected government.