Conversation

Notices

1. Embed this notice
   ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Sunday, 18-May-2025 14:34:15 JST ✧✦Catherine✦✧

   a motherboard from my now-previous laptop doesn't boot; the LED blink sequence decodes to "BIOS flash failure". first of all what the heck? that never happens (but the symptoms when it died were consistent with it happening)

   anyway, it uses plain SPI, and #GlasgowInterfaceExplorer can now analyze SPI transactions very easily

   let's find out

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Sunday, 18-May-2025 14:37:27 JST ✧✦Catherine✦✧

     in reply to

     most likely i'll implement an SPI flash analyzer applet, layered on top of the SPI analyzer; this is a _lot_ of transactions

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Sunday, 18-May-2025 15:56:16 JST ✧✦Catherine✦✧

     in reply to

     update: this is a video BIOS. i went through all that effort and picked the wrong flash IC. fuck.

     (in my defense, there are *at least* four of them in close proximity. i'm not convinced i found every one of them. x86 these days..._

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Sunday, 18-May-2025 16:00:02 JST ✧✦Catherine✦✧

     in reply to

     sir, this is a wendy's^W laptop sku

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 19-May-2025 07:52:27 JST ✧✦Catherine✦✧

     in reply to

     okay, i soldered to the right flash and now i'm getting the expected transactions from the CPU

     and of course, it's using QSPI, so now i need to build a QSPI analyzer applet

     not to worry; i have the power to do this! i even soldered to WP#/HOLD in anticipation

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 19-May-2025 07:52:40 JST ✧✦Catherine✦✧

     in reply to

     • Ronflaix

     @Ronflaix the video BIOS executes in x86 real mode and can check for stuff like that

   • Embed this notice
     Ronflaix (ronflaix@mastodon.gamedev.place)'s status on Monday, 19-May-2025 07:52:42 JST Ronflaix

     in reply to

     @whitequark oh lol I remember seeing that line once. I forgot to plug all of the power connector and the PC was only showing this in text mode, doing nothing else.

     I wonder if the GPU takes over the booting process or blocks it while doing that.

     Edit: hang on, it could just be something on the mobo

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 19-May-2025 10:54:47 JST ✧✦Catherine✦✧

     in reply to

     i've implemented a decoder for SPI communication dumps that can extract not only data, but also SFDP information from communications, meaning you actually learn _which_ exact flash it was just from examining the dump!

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 19-May-2025 11:34:39 JST ✧✦Catherine✦✧

     in reply to

     please enjoy the SPI Flash dump decoder https://github.com/GlasgowEmbedded/glasgow/pull/821

     it even supports 4-byte addressing! no writes (yet) though

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 19-May-2025 11:57:09 JST ✧✦Catherine✦✧

     in reply to

     next step is implementing a QSPI analyzer, to go with the existing SPI analyzer. this is mostly a copy/paste job, i could probably reuse some bits between the two but it's not worth it

   • Embed this notice
     Joel Michael (jpm@aus.social)'s status on Monday, 19-May-2025 12:17:35 JST Joel Michael

     in reply to

     @whitequark … now I’m wondering how to hook up a Glasgow to Wireshark for easy to use protocol decoding.

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 19-May-2025 12:17:35 JST ✧✦Catherine✦✧

     in reply to

     • Joel Michael

     @jpm we actually already have a protocol implementation for https://wiki.wireshark.org/snoop files

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 19-May-2025 13:26:38 JST ✧✦Catherine✦✧

     in reply to

     • Julien Goodwin

     @LapTop006 yeah, quad/octo SPI definitely have commonality

   • Embed this notice
     Julien Goodwin (laptop006@aus.social)'s status on Monday, 19-May-2025 13:26:39 JST Julien Goodwin

     in reply to

     @whitequark although maybe if you also do octospi

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 19-May-2025 13:46:18 JST ✧✦Catherine✦✧

     in reply to

     implemented a QSPI analyzer and verified with pulseview that at least the SPI part of it is functional. (i also have unit tests)

     in the process i discovered that sigrok's SPI flash decoder is completely cooked in how it handles unknown commands. also it doesn't have quad I/O mode at all

     https://github.com/GlasgowEmbedded/glasgow/pull/823

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 19-May-2025 13:56:28 JST ✧✦Catherine✦✧

     in reply to

     • Chuck

     @ChuckMcManis Glasgow also has a very efficient QSPI implementation!

   • Embed this notice
     Chuck (chuckmcmanis@chaos.social)'s status on Monday, 19-May-2025 13:56:30 JST Chuck

     in reply to

     @whitequark Oh that is super awesome! I noticed that Sigrok was falling down here and put my QSPI gateware project back on hold until I could do something like this. Procrastination paying off for me 😁

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 19-May-2025 15:39:09 JST ✧✦Catherine✦✧

     in reply to

     i expanded my 25-series Flash trace analyzer to handle QSPI as well! here it is, decoding EBh Quad I/O Fast Read commands :D :D

   • Embed this notice
     ✧✦Catherine✦✧ (whitequark@mastodon.social)'s status on Monday, 19-May-2025 15:45:16 JST ✧✦Catherine✦✧

     in reply to

     at the output of the trace analyzer, you get a pair of files like this: