| --- Log opened Tue Jan 09 00:00:19 2018 |
| 00:26 | | Kindamoody is now known as Kindamoody[zZz] |
| 00:44 | <@himi> | https://support.google.com/faqs/answer/7625886 |
| 00:44 | <@himi> | Interesting approach to protecting against Spectre |
| 00:45 | <@himi> | Possibly |
| 00:45 | <&McMartin> | Yeah, that one targets the branch predictor variant |
| 00:47 | <@himi> | Essentially emulating the ARM __foo_no_speculate instructions |
| 01:07 | | Derakon[AFK] is now known as Derakon |
| 01:11 | | celticminstrel [celticminst@Nightstar-m9434e.dsl.bell.ca] has joined #code |
| 01:11 | | mode/#code [+o celticminstrel] by ChanServ |
| 01:32 | < [> | it's weird that intel offers the two branch hints but no "don't predict" hint |
| 01:33 | < [> | though i guess adding either hint also works around the issue |
| 01:36 | < [> | err in the path taken case at least. not in the array bound case |
| 02:00 | <&McMartin> | Array bound case it's "load fence" |
| 02:00 | <&McMartin> | You usually use those to implement mutexes but that's the software fix AIUI |
| 02:03 | < [> | what do you think the hw solution will be in future chips |
| 02:05 | < [> | new seperate cache? |
| 02:06 | < [> | basically enable rollback of the cache by keeping a seperate copy that doesn't commit until operations that change it have retired |
| 02:07 | <@himi> | More cache tagging, I suspect, particularly tagging that tracks privilege levels, along with hardware that manages access based on tags |
| 02:07 | <@himi> | So the data is in the cache, but it's only accessible within one zone/domain/something |
| 02:09 | <@himi> | Probably also changes in the management of TLBs - tag the TLB the same way the cache lines are, and then essentially give each zone/domain/whatever a constrained view of cache based on those tags |
| 02:10 | <@himi> | i.e. each virtual memory domain (represented by a set of TLBs that establish a virtual->physical memory map) gets tagged and then on the CPU that tag is used to provide access control |
| 02:13 | < [> | do you know a lot about cpu/mmu design? i'm curious why there isn't a seperate page table and tlb for ring0 |
| 02:14 | < [> | so kernel mem isn't even mapped as far as user mode is concerned |
| 02:14 | <@himi> | Not much about CPU/MMU design |
| 02:14 | <@himi> | It's not that there isn't a separate page table for kernel memory, it's a design choice for performance reasons |
| 02:15 | <@himi> | Each process has a read-only and hidden mapping for kernel memory in its virtual memory space in order to minimise the cost of system calls |
| 02:15 | < [> | it's only hidden in that usermode access to it throws an exception |
| 02:15 | <@himi> | A lot of the work being done to mitigate meltdown and spectre is reversing that design decision |
| 02:15 | < [> | yeah, poor decision in retrospect |
| 02:16 | < Mahal> | It made perfect sense at the time |
| 02:16 | < [> | i'm imagining the kernel table being a superset, like a deque, maybe user ptes grow one way and kernel the other |
| 02:16 | <@himi> | This is one of the reasons for address space randomisation |
| 02:16 | < Mahal> | they so often do~ |
| 02:16 | < [> | then it can scan the table as usual |
| 02:18 | <@himi> | Simply treating the kernel's virtual memory space the same way any other process is treated resolves most of these issues |
| 03:08 | <&McMartin> | I saw something today talking about Apple's response to Meltdown and it noted they were already swapping the kernel out "by hand" already, and it sounded like this was to make the x86 more like the PPC, which *did* have different caches/tables for each ring |
| 03:10 | <@himi> | . . . "swapping the kernel out by hand"? |
| 03:10 | <@himi> | As in, simply having a completely separate virtual address space for the kernel? |
| 03:12 | <&McMartin> | Yeah |
| 03:13 | <&McMartin> | Well |
| 03:13 | <&McMartin> | Mostly |
| 03:13 | <&McMartin> | Much much smaller than Linux or BSD normally did, anyway |
| 03:13 | <&McMartin> | Because the virtual spaces are almost (completely?) disjoint on PPC which of course OSX was first |
| 03:13 | | Vornotron [Vorn@Nightstar-1l3nul.res.rr.com] has joined #code |
| 03:13 | | mode/#code [+qo Vornotron Vornotron] by ChanServ |
| 03:14 | | Vornicus [Vorn@Nightstar-1l3nul.res.rr.com] has quit [Ping timeout: 121 seconds] |
| 03:34 | | macdjord is now known as macdjord|away |
| 04:10 | | * McMartin eyes his math, not completely believing it |
| 04:11 | <&McMartin> | This looks rather suspiciously like I can reblit an entire screen's worth of data in the Game Boy's VBLANK |
| 04:11 | <&McMartin> | That would be a first |
| 04:27 | <~Vornotron> | for the game boy, that ...doesn't really surprise me that much |
| 04:28 | <~Vornotron> | I'm not sure why but it feels like a game that would be able to handle it |
| 04:28 | <~Vornotron> | system |
| 04:43 | | * Alek is reminded of the Pokemon graphical glitches when you do certain things, like hunt MissingNo. |
| 04:44 | <@Alek> | did they store actual game data in the screen data? |
| 04:48 | <~Vornotron> | it's not like the game can tell from the cartridge what it's looking at |
| 04:49 | <~Vornotron> | point it at a thing and say "this is graphics" and it will believe you |
| 05:04 | <&McMartin> | Unless there is cartridge level bankswitching, The tile graphics and tile tables are mapped into RAM at all times and can be accessed with ordinary LD instructions |
| 05:04 | <&McMartin> | However, if you do that while the display needs that data, your memory operation is to a high-impedance connection and the load or store silently fails |
| 05:05 | <&McMartin> | So you want to do this in VBLANK |
| 05:05 | <&McMartin> | Which is only 10 scanlines long, but the LCD's "HBLANK" periods are huuuuuge |
| 05:06 | <&McMartin> | here, have the Ultimate Game Boy Talk: https://www.youtube.com/watch?v=HyzD8pNlpwI |
| 05:07 | <&McMartin> | Which also seems to be the only source for The Good Stuff about how sprites actually get spat out by the PPU |
| 05:07 | | celticminstrel [celticminst@Nightstar-m9434e.dsl.bell.ca] has quit [[NS] Quit: KABOOM! It seems that I have exploded. Please wait while I reinstall the universe.] |
| 05:10 | | Derakon is now known as Derakon[AFK] |
| 05:53 | | macdjord|away is now known as macdjord|slep |
| 05:58 | | himi [sjjf@Nightstar-dm0.2ni.203.150.IP] has quit [Ping timeout: 121 seconds] |
| 06:43 | | Vornlicious [Vorn@Nightstar-m4dupm.sub-174-211-0.myvzw.com] has joined #code |
| 06:47 | | Vorntastic [Vorn@Nightstar-1l3nul.res.rr.com] has quit [Ping timeout: 121 seconds] |
| 07:44 | | Kindamoody[zZz] [Kindamoody@Nightstar-eubaqc.tbcn.telia.com] has quit [Connection closed] |
| 07:45 | | Kindamoody[zZz] [Kindamoody@Nightstar-eubaqc.tbcn.telia.com] has joined #code |
| 07:45 | | mode/#code [+o Kindamoody[zZz]] by ChanServ |
| 08:51 | | himi [sjjf@Nightstar-v37cpe.internode.on.net] has joined #code |
| 08:51 | | mode/#code [+o himi] by ChanServ |
| 09:17 | | Vornotron [Vorn@Nightstar-1l3nul.res.rr.com] has quit [Ping timeout: 121 seconds] |
| 09:48 | <@abudhabi> | Hmmm. |
| 09:49 | <@abudhabi> | Suppose you have domains a.com, b.com, c.com. |
| 09:50 | <@abudhabi> | a.com has a linked website, both being bought from the registrar. |
| 09:50 | < Vornlicious> | Like d.a.com? |
| 09:51 | <@abudhabi> | No, no, the direct translation would be like a "web hotel". As in some form of virtual server that gives viewers a website when they browse to a.com. |
| 09:52 | <@abudhabi> | Is it possible to make it so whenever a.com is browsed to, it redirects to b.com (displaying in the URL bar as b.com) while still giving the same website as it normally does, and whenever c.com is browsed, it redirects to the website under a.com but views b.com in the URL bar? |
| 09:54 | <@TheWatcher> | apache rewrite rules would probably let you do something like that, although the more pertinent question might be "why?" |
| 09:55 | <@abudhabi> | Client. |
| 09:55 | <@abudhabi> | I'm trying to figure out the least painful way of giving them what they want. |
| 09:56 | <&[R]> | So they want to have b.com be non-deterministic in what it displays when you access it? |
| 09:56 | <&[R]> | Why do they think this is a good idea? |
| 09:57 | <@abudhabi> | The goal of this exercise is to rename a.com to b.com, routing everything to the pre-existing website. |
| 09:57 | <@abudhabi> | Including the previous a.com, for legacy reasons. |
| 09:57 | <&[R]> | So just clone a.com to be b.com, and make a.com redirect? |
| 09:59 | <@abudhabi> | I'll look into that. |
| 10:06 | | Kindamoody[zZz] is now known as Kindamoody |
| 14:21 | <@abudhabi> | Hmm. How difficult is to forward mail from one domain to another, preferably with DNS records? |
| 14:22 | <@abudhabi> | In this instance, I would want foo@bar.com auto-redirected to foo@bar.org. |
| 14:22 | <&[R]> | Can't be done purely with DNS |
| 14:22 | <@abudhabi> | Right, so it needs a mail server under the domain. |
| 14:22 | <&[R]> | Email server has to be configured to make the old domain emails go to new domain |
| 14:22 | <@abudhabi> | Easy, hard? |
| 14:22 | <&[R]> | Not really, they can have the same SMTP server |
| 14:23 | <@abudhabi> | But the domain records need to point at that SMTP server, yes? |
| 14:23 | <&[R]> | Easy with some SMTP servers, I would suspect MS complicates that unnecessarily. |
| 14:23 | <&[R]> | Yes |
| 14:23 | <@abudhabi> | OK. Sounds easy enough, then. |
| 14:24 | <&[R]> | You might also need to transfer some other stuff, like DPKI (sp?) and SPF |
| 14:28 | <@abudhabi> | What do these do? |
| 14:29 | <&[R]> | SPF's a DNS record that tells email servers which email servers are allowed to send email for a domain. |
| 14:29 | <&[R]> | It's an anti-spoofing thing |
| 14:36 | < Vornlicious> | Dkim |
| 14:37 | <@abudhabi> | Hmm. Can you make web traffic to a domain get redirected to another domain, but leave the mail server directions intact? |
| 14:37 | < Vornlicious> | Yes. |
| 14:38 | < Vornlicious> | Typically you run a big forwarding bot on http and https |
| 14:38 | <@abudhabi> | Cool! Just to clarify, I want to gradually move to another domain here. HTTP and stuff goes to the new domain, but mail stays put for the next step. |
| 14:38 | <@abudhabi> | OK, yeah, I figure that would help, being a non-DNS issue. |
| 14:38 | < Vornlicious> | This actually makes the domain appear to move for web but it does nothing whatsoever for non-web stuff |
| 14:49 | | Degi [Degi@Nightstar-4puh6e.dyn.telefonica.de] has joined #code |
| 14:49 | | Vornlicious [Vorn@Nightstar-m4dupm.sub-174-211-0.myvzw.com] has quit [Connection closed] |
| 14:50 | <&[R]> | It might be useful to know how DNS works in this situation actually. DNS is simply a name to number mapping system. So a single DNS name goes to a single or set of IP addresses. |
| 14:51 | <&[R]> | DNS doesn't do anything with ports, and therefore can't do anything about sending different services to different servers. However, there are some things to note about that. |
| 14:52 | <&[R]> | SMTP gets its own DNS records, which means they can go to an entirely different server. Additionally, you can use IPv6 during DNS name resolution, which can let you send IPv4 and IPv6 traffic to different hosts if you wanted. |
| 14:53 | <&[R]> | Most protocols don't know anything about what DNS name a client requested them on. The two big exceptions are HTTP(S) and SMTP, which have special fields to share that information. |
| 14:54 | | Vorntastic [Vorn@Nightstar-m4dupm.sub-174-211-0.myvzw.com] has joined #code |
| 15:21 | | Vorntastic [Vorn@Nightstar-m4dupm.sub-174-211-0.myvzw.com] has quit [[NS] Quit: Bye] |
| 15:21 | | Vorntastic [Vorn@Nightstar-1l3nul.res.rr.com] has joined #code |
| 15:26 | | crystalclaw [crystalclaw@Nightstar-12q9ui.xyz] has quit [Ping timeout: 121 seconds] |
| 15:27 | | crystalclaw [crystalclaw@Nightstar-12q9ui.xyz] has joined #code |
| 15:27 | | mode/#code [+o crystalclaw] by ChanServ |
| 15:27 | | Vornicus [Vorn@Nightstar-1l3nul.res.rr.com] has joined #code |
| 15:27 | | mode/#code [+qo Vornicus Vornicus] by ChanServ |
| 16:10 | | Vornotron [Vorn@Nightstar-1l3nul.res.rr.com] has joined #code |
| 16:10 | | mode/#code [+qo Vornotron Vornotron] by ChanServ |
| 16:11 | | Degi [Degi@Nightstar-4puh6e.dyn.telefonica.de] has quit [Connection closed] |
| 16:13 | | Vornicus [Vorn@Nightstar-1l3nul.res.rr.com] has quit [Ping timeout: 121 seconds] |
| 17:13 | | Kindamoody is now known as Kindamoody|afk |
| 17:43 | | Kindamoody|afk [Kindamoody@Nightstar-eubaqc.tbcn.telia.com] has quit [Ping timeout: 121 seconds] |
| 17:56 | | gnolam_ [lenin@Nightstar-ego6cb.cust.bahnhof.se] has joined #code |
| 17:58 | | gnolam [lenin@Nightstar-ego6cb.cust.bahnhof.se] has quit [Ping timeout: 121 seconds] |
| 18:00 | | gnolam_ is now known as gnolam |
| 18:00 | | mode/#code [+o gnolam] by ChanServ |
| 18:15 | | Alek [Alek@Nightstar-7or629.il.comcast.net] has quit [Ping timeout: 121 seconds] |
| 18:19 | | Alek [Alek@Nightstar-7or629.il.comcast.net] has joined #code |
| 18:19 | | mode/#code [+o Alek] by ChanServ |
| 18:25 | | Reiv [NSkiwiirc@Nightstar-ih0uis.global-gateway.net.nz] has quit [[NS] Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] |
| 18:27 | <@abudhabi> | [R]: Right, right, I know the basics. |
| 18:27 | <@abudhabi> | Pity that the person the client is replacing me (in-house), knows less than I do. |
| 18:28 | | Reiv [NSkiwiirc@Nightstar-ih0uis.global-gateway.net.nz] has joined #code |
| 18:28 | | mode/#code [+o Reiv] by ChanServ |
| 18:29 | <@abudhabi> | I was a little concerned and confused why they were apparently aiming to do something that would essentially break the company website and email, because the old-timey-technical boss wants to do the equivalent of sorting files into folders according to their extension. |
| 18:32 | | Degi [Degi@Nightstar-4puh6e.dyn.telefonica.de] has joined #code |
| 19:06 | | Kindamoody|autojoin [Kindamoody@Nightstar-v552ts.tbcn.telia.com] has joined #code |
| 19:06 | | mode/#code [+o Kindamoody|autojoin] by ChanServ |
| 19:08 | | macdjord|slep is now known as macdjord |
| 19:28 | | Degi [Degi@Nightstar-4puh6e.dyn.telefonica.de] has quit [[NS] Quit: Leaving] |
| 19:43 | | Kindamoody|autojoin [Kindamoody@Nightstar-v552ts.tbcn.telia.com] has quit [Ping timeout: 121 seconds] |
| 19:53 | | crystalclaw [crystalclaw@Nightstar-12q9ui.xyz] has quit [Ping timeout: 121 seconds] |
| 19:57 | | crystalclaw [crystalclaw@Nightstar-12q9ui.xyz] has joined #code |
| 19:58 | | mode/#code [+o crystalclaw] by ChanServ |
| 20:23 | | Reiv [NSkiwiirc@Nightstar-ih0uis.global-gateway.net.nz] has quit [[NS] Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] |
| 20:24 | | himi [sjjf@Nightstar-v37cpe.internode.on.net] has quit [Ping timeout: 121 seconds] |
| 20:25 | | Reiv [NSkiwiirc@Nightstar-ih0uis.global-gateway.net.nz] has joined #code |
| 20:25 | | mode/#code [+o Reiv] by ChanServ |
| 20:29 | < Mahal> | https://bugs.chromium.org/p/project-zero/issues/detail?id=1272 |
| 20:32 | | Reiv [NSkiwiirc@Nightstar-ih0uis.global-gateway.net.nz] has quit [[NS] Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] |
| 20:34 | | Reiv [NSkiwiirc@Nightstar-ih0uis.global-gateway.net.nz] has joined #code |
| 20:34 | | mode/#code [+o Reiv] by ChanServ |
| 20:55 | <@TheWatcher> | Status: fixed. |
| 20:55 | <@TheWatcher> | ahahaahahahahahahahahahhahahahahha |
| 20:56 | <@TheWatcher> | -_- |
| 20:56 | <@TheWatcher> | More accurate status: https://media.giphy.com/media/26FPy3QZQqGtDcrja/giphy.gif |
| 21:04 | <&McMartin> | Look, RESOLVED FIXED just means the commit has hit the repository and *maybe* that it's been tested. |
| 21:04 | <&McMartin> | (Assuming "It passed the tests" isn't CLOSED) |
| 21:32 | | Reiv [NSkiwiirc@Nightstar-ih0uis.global-gateway.net.nz] has quit [[NS] Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] |
| 21:33 | | Reiv [NSkiwiirc@Nightstar-ih0uis.global-gateway.net.nz] has joined #code |
| 21:33 | | mode/#code [+o Reiv] by ChanServ |
| 21:50 | <&[R]> | For fucks sake |
| 21:50 | <&[R]> | Outlook fucking corrupted its .ost file |
| 22:01 | | Reiv [NSkiwiirc@Nightstar-ih0uis.global-gateway.net.nz] has quit [[NS] Quit: http://www.kiwiirc.com/ - A hand crafted IRC client] |
| 22:05 | | himi [sjjf@Nightstar-dm0.2ni.203.150.IP] has joined #code |
| 22:05 | | mode/#code [+o himi] by ChanServ |
| 22:17 | <@TheWatcher> | [R]: woe :( |
| 22:29 | | Kindamoody|autojoin [Kindamoody@Nightstar-eubaqc.tbcn.telia.com] has joined #code |
| 22:29 | | mode/#code [+o Kindamoody|autojoin] by ChanServ |
| 22:34 | | Kindamoody|autojoin is now known as Kindamoody |
| 22:42 | | crystalclaw [crystalclaw@Nightstar-12q9ui.xyz] has quit [Ping timeout: 121 seconds] |
| 22:43 | | crystalclaw [crystalclaw@Nightstar-12q9ui.xyz] has joined #code |
| 22:43 | | mode/#code [+o crystalclaw] by ChanServ |
| 23:13 | | Kindamoody [Kindamoody@Nightstar-eubaqc.tbcn.telia.com] has quit [Ping timeout: 121 seconds] |
| 23:26 | | Reiv [NSkiwiirc@Nightstar-ih0uis.global-gateway.net.nz] has joined #code |
| 23:26 | | mode/#code [+o Reiv] by ChanServ |
| 23:31 | | [ [art@Nightstar-lbl.59v.61.68.IP] has quit [Ping timeout: 121 seconds] |
| 23:55 | | [ [art@Nightstar-lbl.59v.61.68.IP] has joined #code |
| --- Log closed Wed Jan 10 00:00:20 2018 |