- Talk - lists.gtalug.org

growing an EFI System Partition (ESP)
by D. Hugh Redelmeier 08 Nov '23

08 Nov '23

(This is a recipe. If you find errors or think of improvements, please let me know.) == context == The ESP is the partition to hold anything the UEFI firmware can run. Mostly bootloaders but also diagnostics and firmware updates. It is a FAT filesystem in a partition with a special type. In a running Linux, it is usually mounted on /boot/efi. This stuff is a little bit tricky and not well-documented. I tried to be very detailed to make sure you don't get lost. == motivation == The ESP doesn't have to be very big, but sometimes it isn't big enough. I often buy computers pre-loaded with Windows -- typically cheaper than computers without Windows. Sometimes the ESP is 100MB. This has pinched. So I want to increase the size. I have a feeling that 200MB would be a good size, but I will probably settle on 400MB. == naive method and a failure == Tools: - a live USB Linux system. [You don't want to be doing partition surgery on the disk from which you are running.] - some place to temporarily hold the contents of the ESP. It could be on another partition of the same disk or on a USB devicee. - gparted for partition manipulation. You could substitute with something like parted(8). - gnome "disks" (gnome-disk-utility). I find it convenient for mounting, unmounting, and looking at the properties of disks and partition. You can choose to do these things with other tools. 1. boot the live Linux stick. 2. if your live system doesn't include gparted, install it. Fedora: "sudo dnf install gparted" debian or Ubuntu: "sudo apt install gparted" 3. run gparted and point it at the disk with the too-small ESP. Make sure that none of that disk's partitions is mounted. 3. shift and shrink partitions to make sufficient unallocated space adjacent to the ESP. If you are going to install Linux on the disk, this would be a good time to make space for it too. Make sure to tell gparted to "apply" these changes. 4. If you changed the size of any NTFS partitions, gparted might have damaged some redundant metadata. Fix it now. - boot into Windows - ask it to scan and repair the volume(s) - reboot into your live Linux system and get back into gparted 5. ask gparted to resize the ESP. It should succeed in resizing the partition but may fail in resizing the FAT filesystem. The failure message will be rather obscure: No Implementation: GNU Parted cannot resize this partition to this size. We're working on it! What it means is that libparted-fs-resize doesn't have code to deal with some common forms of FAT. The message is wrong: the partition has been resized but the filesystem it contains has not. Oh, and nobody has worked on this in a decade. If you had no failure, STOP HERE, you are done. Continuing past FAT filesystem resize failure: 1. exit gparted 2. mount the old ESP. GNOME "disks" has a button for that. 3. copy the contents of the ESP somewhere safe. It could be to a USB device or even a non-ESP partition on the system disk. "sudo cp -a ESP SAVED" ESP is the ESP mount point. SAVED is to be a new directory in the place where you have enough free space 4. unmount the old ESP 5. Make a new, larger FAT filesystem, filling the partition: "sudo mkfs.vfat X" where X is the device name of the ESP partition. On my system, it was /dev/nvme0n1p1 but yours might well be different This will wipe the old FAT filesystem and make an empty new one that fills the partition. 6. mount the new ESP 7. copy the old contents of the ESP into the new ESP. "sudo cp -a SAVED/* ESP Note: the /* bit is to avoid creating an extra directory layer in the ESP. * isn't great but FAT filesystems don't have files starting with ".". 8. shut down Linux; unplug the live stick FINI

2 2

"RISC-V technology emerges as battleground in US-China tech war"
by D. Hugh Redelmeier 06 Nov '23

06 Nov '23

<https://www.reuters.com/technology/us-china-tech-war-risc-v-chip-technology…> Grrr. Some US lawmakers want to restrict US companies from working on RISC-V. They are worried it will benefit China (of course it will, and the rest of us too). It is true that ARM's ownership has allowed the US government to make things very difficult for Huawei. No wonder China likes RISC-V.

4 7

(very off topic) where to buy old AM/FM receiver?
by William Park 31 Oct '23

31 Oct '23

Hi, a very off topic... While cleaning my house, I found box speakers and LP record player. Brand name "Sanyo". What's missing is AM/FM receiver that came as a set. Do you know where I can buy an old stereo AM/FM receiver? --

9 13

Old (mostly) O'Reilly books available free
by Giles Orr 30 Oct '23

30 Oct '23

I have a number of O'Reilly books that I haven't used in years (and a couple other tech titles as well). It's time I got rid of them, and it seems a waste to toss them in the recycling. If any of you are interested in these, pick up will be in North York: either North York Central Library or nearby. Please see the image below for the titles - and note that if there are modern versions of these titles, the copies you see here are almost certainly NOT the most recent version. https://gilesorr.com/forOthers/books.jpg -- Giles https://www.gilesorr.com/ gilesorr(a)gmail.com

2 1

No keyboard at boot?
by Peter King 28 Oct '23

28 Oct '23

The saga continues. While waiting for a replacement motherboard for my Lenovo T5, I hauled out a computer I had taken out of service about 16 months ago to have a working machine until I finally can install the motherboard. But here's the thing. When it boots up, I can use the keyboard and mouse at the BIOS screen and through most of the process -- but near the end, after enabling the network fails, it loses the keyboard. The mouse stays connected. So I get a login screen but no way to talk to it. Same problem using other keyboards. Eventually I figured out how to boot from a USB stick, and that gets all the way up with the keyboard, so it's something odd in the setup on the computer itself. I haven't tried to ssh in, because at the moment it tries to boot into a static ethernet address which isn't available, and so network connectivity fails. The system runs Arch Linux. The "easy" thing to do is just re-install Arch and start with a clean slate, which is likely what I'll do, but if anyone knows a quick and dirty fix I'd at least be able to save some configuration information from the computer. (Yes, I can do that from an Arch install USB by mounting / chrooting into it, but I'd rather understand the problem first.) Any idea what gives? Obviously a module for running the keyboard isn't being loaded somewhere...

5 12

why I like shared libraries -- no longer a popular position
by D. Hugh Redelmeier 27 Oct '23

27 Oct '23

<https://arstechnica.com/security/2023/09/incomplete-disclosures-by-apple-an…> A bug was found (painfully -- a zero day) in Apple's Safari and (separately) in Google's Chrome. This is a pretty serious bug -- it was used to spy on an opposition politician in Egypt. It is the same bug, and this was not reported. It turns out that the bug is in libwebp. "WebP codec is a library to encode and decode images in WebP format." libwebp is used in a lot of programs. On my Fedora 38 system, it is a shared library so it can be fixed in one update. Except where the library is copied (for example, statically linked, or used in a container of some sort). Electron is one thing that requires copies and the article lists a lot of applications built on Electron What a mess. What a mistake.

7 11

intel etherexpress pro 100+ safe driver download?
by Karen Lewellen 25 Oct '23

25 Oct '23

Hi everyone, My dream machine is being constructed. Thanks to the members here connecting with James. There is indeed a DOS packet driver included with the set Intel provides for the etherexpress pro 100+ Trick seems to be finding a safe source, with my having access to the way of putting things on floppies. For example, I found an option form archive.org, that included a 1.57 .img file. Slightly too large for the 1.44 disks I sought to zip..only for pk zip to suggest the file may have a problem. Any ideas? Thanks profoundly, Kare

2 1

November Meeting Notice
by Alan Heighway 22 Oct '23

22 Oct '23

Hello GTALUG, Due to a misunderstanding we were unable to hold the AGM in October as planned. However we are on track to do so at our November meeting, followed by a talk by our own Dave Collier-Brown titled "You Don’t know Jack about Application Performance." Our AGM proceedings will start with a treasurer's report, then a vote on the merger CLUE, and conclude with filling vacant board positions (with an election if required). Only paid GTALUG members can participate in the voting. This is a perfect time to sign up or renew your membership. To do so send an Interac e-transfer to membership(a)gtalug.org for $20 with your Name and email address. That's it, you are a voting member! Thank you, and I hope to see you at our next meeting. Alan. *A*lan *H*eighway <http://heighway.ca/> heighway.ca VA3WAH / VA3YKZ [image: Please consider the environment before printing]

2 1

sigh, hardware resources?
by Karen Lewellen 21 Oct '23

21 Oct '23

Hi everyone, will ask the question simply, as it is not one I have needed to ask for a good decade. Need a resource for older computer hardware, freecycle was terrific for this sort of thing, as was the talent who would build mine. Unfortunately a visit by someone from renew computer technology Ontario, www.rcto.org damaged my main computer, at least the motherboard, possibly the hard drive. I have a bandaged machine, does not boot to the c drive, using a floppy, with the second drive from my main placed in this unit. Now some of my critical synthesizer files are being damaged as well. screen reader utilities..so far at least. While I do have talent, quality talent to help me with the construction, finding say a p3 or p4 may be the challenge. Any ideas? Cheers, Karen

8 22

CVE-2023-43641: libcue bug allows Remote Code Execution in (at least) GNOME
by D. Hugh Redelmeier 20 Oct '23

20 Oct '23

TL;DR: update libcue to a version released after October 10. I read this last night: <https://arstechnica.com/information-technology/2023/10/one-click-remote-cod…> <https://nvd.nist.gov/vuln/detail/CVE-2023-43641> Summary: - libcue has a bug that allows an attacker to execute arbitrary machine code - libcue handles CUE sheet files - GNOME (at least) uses libcue for file indexing when it finds a suitable file - a bad actor can give you a bad CUE file without you knowing it (eg. via browsing a web site) - GNOME's tracker will automatically run libcue on the CUE file -- kaboom. You can update your system to get a fixed libcue. The fix is in version 2.3.0. Fedora 37 and 38 have fixes in version 2.2.1-13. I checked this by using "rpm -q --changelog libcue | less".

2 1