From: Digestifier To: Linux-Development@senator-bedfellow.mit.edu Reply-To: Linux-Development@senator-bedfellow.mit.edu Date: Mon, 10 Oct 94 15:13:11 EDT Subject: Linux-Development Digest #287 Linux-Development Digest #287, Volume #2 Mon, 10 Oct 94 15:13:11 EDT Contents: Hardware PROBLEM HELP! (Axel Winter) Re: 3c503 problem (Greg Bruell) Re: Beautifying Linux/Xfree (Oliver Mai) Re: Improving SLIP latency under Linux (Rob Janssen) tty bug and fix (for 1.1.51) (Frank Lofaro) Re: Improving SLIP latency under Linux (Rob Janssen) Re: Question: Any DOS/4GW? (Rob Janssen) executable size too big (Ramana Ramachandran) tty bug and fix (for 1.1.51) (Frank Lofaro) Re: ext2fs vs. Berkeley FFS (Hugh Strong) Re: BSD/386 vs. Linux Performance (Alan Cox) ---------------------------------------------------------------------------- From: wintera@morakot.nectec.or.th (Axel Winter) Subject: Hardware PROBLEM HELP! Date: 8 Oct 1994 04:17:33 GMT Hallo, I really have a problem with my hardware ... If I touch any place of the system or a conected part (priter, termial) at a place where is steel, i get a eletric shock. Next to that sometimes the keyboard stopps working at all ... The system isn't halted or jump off ... it's only the keyboard which is off line (mouse, terminals etc. works fine) ... May be this is an effect from electric ... The system is operated by 220V here in Thailand. Hope some can help. Axel ------------------------------ From: gbruell@wellfleet.com (Greg Bruell) Crossposted-To: comp.os.linux.help Subject: Re: 3c503 problem Date: 10 Oct 1994 15:41:05 GMT OK. I've done some more homework. Based on looking at net/inet/dev.c to find where /proc/net/dev is created I found the stat rx_errors. I grepped for that and found it in several places but the only place in the 3c503 driver was actually in the file 8390.c. Here's the code: /* Check for bogosity warned by 3c503 book: the status byte is n ever written. This happened a lot during testing! This code shoul d be cleaned up someday. */ if (rx_frame.next != next_frame && rx_frame.next != next_frame + 1 && rx_frame.next != next_frame - num_rx_pages && rx_frame.next != next_frame + 1 - num_rx_pages) { ei_local->current_page = rxing_page; outb(ei_local->current_page-1, e8390_base+EN0_BOUNDARY); ei_local->stat.rx_errors++; continue; } The stat is written only in drivers. It is read in only one place in the kernel. Based on a suggestion someone sent me I tried programmed IO mode. This worked better but not reliably. I'm not shadowing memory according to my BIOS setup but the idea that this is related to memory/IO conflicts seems plausible. One last thing is that ARP works even though nothing else does. Gregory Bruell Wellfleet Communication, Inc. Donald Becker (becker@cesdis.gsfc.nasa.gov) wrote: : In article <376oda$55p@paperboy.wellfleet.com>, : Greg Bruell wrote: : >I'm seeing the same problem with: : > 1.1.51, P60 : > : >I've checked /var/adm/notice and /proc/net/dev and there is no : >further information. I gave a quick glance at the driver and it : >doesn't look like it logs much. I'm using the internal transciever. : >The card and drop have been independently verified as functional. : >Is anyone else seeing this with 1.1.?? ? : What does /proc/net/dev report? The 'rx-errors' reported by ifconfig is a : catch-all category. Most errors also cause one of the other error counters : to increment. : -- : Donald Becker becker@cesdis.gsfc.nasa.gov : USRA-CESDIS, Center of Excellence in Space Data and Information Sciences. : Code 930.5, Goddard Space Flight Center, Greenbelt, MD. 20771 : 301-286-0882 http://cesdis.gsfc.nasa.gov/pub/people/becker/whoiam.html ------------------------------ From: mai@x4u2.desy.de (Oliver Mai) Crossposted-To: comp.os.linux.misc Subject: Re: Beautifying Linux/Xfree Date: 10 Oct 1994 10:47:08 GMT ... Eric Jeschke wrote: > If there was a default "standard" GUI available on most distributions > it would be possible to write a introduction to Linux (a la Welch) > from a GUI perspective. Whether this is a good idea or not depends on > your ideas about how to teach Unix. IMHO, I think you can wean newbies > off of the mainstream PC OSes more easily with a good GUI and then let > them gradually cut their teeth with more and more command-line. Agreed! > It might be a good idea to make a > GUI "mini-distribution" (package) that just contains a standardized > GUI setup (GREAT + fvwm + customized *rc files + selected GUI apps). I used to run GREAT for some time. But firstly one needs Motif to run GREAT with acceptable performance, and secondly I think fvwm and xfm together give at least the functionality of GREAT without requiring as much resources. For example the GREAT filemanager might look better than xfm, but xfm is more powerful. The xfm file and application managers support drag and drop much better than GREAT. E.g. one can drag files or directories directly into the application manager, then drop files onto the icons of e.g. executables or directories in either the application or file manager, and so on. Once one has a well preconfigured xfm and fvwm one has a very powerful desktop, which IMO is superior to MS Windows', because the xfm application manager is much more flexible than Windows' program manager. Drawbacks are: 1) fvwm does not support drag and drop on desktop icons (in contrast to Windows, but, IMO drag and drop in the application manager is better in practice). 2) There can be only one instance of the xfm application manager. 3) xfm is too slow (esp. changing to directories with many files), but not compared to GREAT's filemanager 4) xfm can only do one operation a time, so if you copy a file in one xfm filemanager window, the other xfm windows are blocked. I would advocate a GUI package with well preconfigured xfm and fvwm. Oliver Mai ------------------------------ From: rob@pe1chl.ampr.org (Rob Janssen) Subject: Re: Improving SLIP latency under Linux Reply-To: pe1chl@rabo.nl Date: Mon, 10 Oct 1994 08:52:41 GMT According to Al Longyear: > For this to work the best, you need to implement IP_TOS on both sides > of the link. If you are doing an ftp transfer to read a large file, > then it the remote side (the one running ftpd) which is doing the > majority of the transmission. Their frames need to have a background > (7) priority. Note in my application Linux on one side is used only as a router. I cannot change the ftpd. So, if it does not use another than the default TOS, I will not have this advantage when using TOS based queueing. Also, I don't really understand the "background (7) priority". I think there are 8 priorities encoded in the TOS and 7 is the HIGHEST of them. Besides, there are the bits that define optimization of one specific aspect of the service. They can be used in the equation as well. Unfortunately, existing systems send a TOS of 00, which is the lowest priority and no specific optimization. It will be difficult to assign relative priorities to some traffic which are lower than this default TOS. (unless you assign it some special meaning) > There have been kludges to support the queueing in the PPP or SLIP > drivers. They look at the port numbers being used and determine a > "priority" based upon the list of known interactive ports. THIS IS A > KLUDGE. It is only because the TCP/IP software did not allow the > application to specify the proper value. > > (Morningstar PPP/SLIP and the BSD pppd package are two that I am > familiar with doing this kludge.) I already mentioned it in my post. KA9Q is another package which uses it. It is a kludge, but it works well. > >Peeking in the TCP headers inside the IP routing code (checking for port > >numbers, so that 20 and 25 can get a lower priority than 23 even when the > >TOS is the same) seems to be the best solution... (although dirty) > > Why bother? It is done in all of the daemons and client code now in > place for Linux. That only helps when the client and server are on Linux machines. This isn't the case in my environment. Some of them are on SYSV.4 and some on Novell machines, which are unchangable. (no source, no knowledgable manufacturer) > However, as the saying goes, "the proof is in the pudding". Please > feel free to make the changes that you see fit to your copy of the > kernel and try it out. I would be interested in hearing about your > results. > > I do ask one favor (or condition if you wish). That is that you use > the current kernel 1.1.52+ and the latest versions of all of the > networking software. (If you use my favorite ftp client, ncftp, then > you need a fairly new one which supports the IP_TOS socket option.) We are now running 1.1.50 and the utilities from slackware 2.0 I hope to upgrade soon, but I often find it difficult to locate recent versions of the networking binaries... (kernel is no problem) > >In fact I think the "always use a big window because that performs so > >nicely on fast links" may be not optimal. One could argue that the optimum > >window size for a TCP is such that it can just keep sending (i.e. an > >ACK comes back just when the sending TCP runs out of send window), and > >not much larger than that. The sending TCP could adapt its send window > >accordingly (staying below the negotiated window size). > > That is the slow start algorithm presently in place. I think the slow start algorithm normally does not do this. It increases the window as long as no negative feedback (packet loss or QUENCH) comes back, without looking at the amount of data in the send window. It may be that the Linux implementation does this differently, but unfortunately the structure and commenting of the code makes it difficult to say for sure (I know KA9Q TCP well, but I am not familiar with the Linux TCP code). What I propose is to clip the send window at the point where there is just enough window to keep it from stop-and-wait behaviour. Is this currently being done? It would be helpful when there would be more of the TCP state visible for the user, e.g. a detailed netstat for one session. KA9Q provides this, and it can give a lot of insight in what the different algorithms are really doing... Rob -- ========================================================================= | Rob Janssen | AMPRnet: rob@pe1chl.ampr.org | | e-mail: pe1chl@rabo.nl | AX.25 BBS: PE1CHL@PI8UTR.#UTR.NLD.EU | ========================================================================= ------------------------------ From: ftlofaro@unlv.edu (Frank Lofaro) Subject: tty bug and fix (for 1.1.51) Date: Wed, 5 Oct 94 04:57:21 GMT There was a bug in the tty code involving EOF and EOF characters that was fixed a while ago, but the fix has become ineffective, and the bug has returned. The bug involves the fact a false EOF is returned when the EOF character is entered with characters waiting (i.e. the EOF character is meant as a push, not as a real EOF). The latest kernel 1.1.51 exhibits this bug. To see the bug in action: type dd bs=4 then enter any multiple of 4 characters, ctrl-d, and the dd will terminate with an end of file. A lot of other un*xes screw up on this too, but Linux had it fixed, and it was not too hard, and this patch is not too difficult, and fixes the problem (and does not appear to create any new bugs; I have tested it in cooked and cbreak modes, with zero and non-zero min and time values it all seems okay). Hopefully this (or a better, more elegant fix) will go into 1.2.0: (I removed the old fix, since it does not do any good anymore) diff -r -u -N linux.dist/drivers/char/n_tty.c linux/drivers/char/n_tty.c --- linux.dist/drivers/char/n_tty.c Sun Sep 25 00:41:14 1994 +++ linux/drivers/char/n_tty.c Sun Sep 25 21:44:35 1994 @@ -453,6 +453,8 @@ goto handle_newline; } if (c == EOF_CHAR(tty)) { + if (tty->canon_head != tty->read_head) + set_bit(TTY_PUSH, &tty->flags); c = __DISABLED_CHAR; goto handle_newline; } @@ -718,24 +720,6 @@ *nr -= n; } -/* - * Called to gobble up an immediately following EOF when there is no - * more room in buf (this can happen if the user "pushes" some - * characters using ^D). This prevents the next read() from falsely - * returning EOF. - */ -static inline void gobble_eof(struct tty_struct *tty) -{ - cli(); - if ((tty->read_cnt) && - (tty->read_buf[tty->read_tail] == __DISABLED_CHAR) && - clear_bit(tty->read_tail, &tty->read_flags)) { - tty->read_tail = (tty->read_tail+1) & (N_TTY_BUF_SIZE-1); - tty->read_cnt--; - } - sti(); -} - static int read_chan(struct tty_struct *tty, struct file *file, unsigned char *buf, unsigned int nr) { @@ -744,6 +728,9 @@ unsigned char *b = buf; int minimum, time; int retval = 0; + int size; + +do_it_again: if (!tty->read_buf) { printk("n_tty_read_chan: called with read_buf == NULL?!?\n"); @@ -858,7 +845,6 @@ put_fs_byte(c, b++); if (--nr) continue; - gobble_eof(tty); break; } if (--tty->canon_data < 0) { @@ -896,7 +882,14 @@ current->state = TASK_RUNNING; current->timeout = 0; - return (b - buf) ? b - buf : retval; + size = b - buf; + if (size && nr) + clear_bit(TTY_PUSH, &tty->flags); + if (!size && clear_bit(TTY_PUSH, &tty->flags)) + goto do_it_again; + if (!size && !retval) + clear_bit(TTY_PUSH, &tty->flags); + return (size ? size : retval); } static int write_chan(struct tty_struct * tty, struct file * file, diff -r -u -N linux.dist/include/linux/tty.h linux/include/linux/tty.h --- linux.dist/include/linux/tty.h Wed Aug 10 09:26:44 1994 +++ linux/include/linux/tty.h Wed Aug 10 09:26:44 1994 @@ -247,6 +247,7 @@ #define TTY_EXCLUSIVE 3 #define TTY_DEBUG 4 #define TTY_DO_WRITE_WAKEUP 5 +#define TTY_PUSH 6 #define TTY_WRITE_FLUSH(tty) tty_write_flush((tty)) ------------------------------ From: rob@pe1chl.ampr.org (Rob Janssen) Subject: Re: Improving SLIP latency under Linux Reply-To: pe1chl@rabo.nl Date: Mon, 10 Oct 1994 09:00:27 GMT In <37981n$9bs@gate.noris.de> urlichs@smurf.noris.de (Matthias Urlichs) writes: >> >3- turn off error correction in the modems >> I tried this to no avail. >Modems should have a configuration option to also turn off (or tune) the >RTS threshold values for the modem's send buffer. Unfortunately, in most >modems there is no such thing. We could see more sensible modems if more >people would complain. Some high-end modems have the option to select between "large" and "small" buffers, but what is meant by "small" is still a bit large for this purpose. It would be nice when the modem itself would do the priority-queueing. Of course one cannot expect this from a modem with a serial interface, but what we could have is a modem with an ethernet connection which operates at the packet level and could do simple priority decisions. With the ever increasing speed of modems, I think an ethernet interface would have many advantages. (like higher datarate, cheap interface to many systems, no multiport serial cards needed to build a modempool, easier to control and monitor the modem state without using in-band (Hayes) signalling, etc) Unfortunately, it seems that modem manufacturers bet on souped-up serial port cards and/or PC-specific parallel interfaces instead. :-( Rob -- ========================================================================= | Rob Janssen | AMPRnet: rob@pe1chl.ampr.org | | e-mail: pe1chl@rabo.nl | AX.25 BBS: PE1CHL@PI8UTR.#UTR.NLD.EU | ========================================================================= ------------------------------ From: rob@pe1chl.ampr.org (Rob Janssen) Subject: Re: Question: Any DOS/4GW? Reply-To: pe1chl@rabo.nl Date: Mon, 10 Oct 1994 09:06:11 GMT In <37aebq$2fj@manuel.anu.edu.au> wlee@clubX.anu.edu.au (Wally) writes: >Is there a DOS/4GW for Linux? or its emulator? >Yes?! Where can I get it? >No?! Anyone working on it? No, it is not needed. Linux can have 3GB in each user process without using such hacks. Rob -- ========================================================================= | Rob Janssen | AMPRnet: rob@pe1chl.ampr.org | | e-mail: pe1chl@rabo.nl | AX.25 BBS: PE1CHL@PI8UTR.#UTR.NLD.EU | ========================================================================= ------------------------------ From: ramana@pen320.lexington.ibm.com (Ramana Ramachandran) Subject: executable size too big Date: 8 Oct 1994 21:06:53 GMT Reply-To: ramana@vnet.ibm.com Hi I recently upgrded my kernel from 1.0.9 to 1.1.45 ( along with it I upgraded gcc 2.4.5 to 2.5.8, libc from 4.0.4 to 4.5.26 ) and now when I compile stuff the executable are about 3 times the size of the original executables. How do I find if I am linking the dynamic libraries or not? Any help would be greatly appreciated. Thanks ramana ------------------------------ From: ftlofaro@unlv.edu (Frank Lofaro) Subject: tty bug and fix (for 1.1.51) Date: Wed, 5 Oct 94 04:41:29 GMT There was a bug in the tty code involving EOF and EOF characters that was fixed a while ago, but the fix has become ineffective, and the bug has returned. The bug involves the fact a false EOF is returned when the EOF character is entered with characters waiting (i.e. the EOF character is meant as a push, not as a real EOF). The latest kernel 1.1.51 exhibits this bug. To see the bug in action: type dd bs=4 then enter any multiple of 4 characters, ctrl-d, and the dd will terminate with an end of file. A lot of other un*xes screw up on this too, but Linux had it fixed, and it was not too hard, and this patch is not too difficult, and fixes the problem (and does not appear to create any new bugs; I have tested it in cooked and cbreak modes, with zero and non-zero min and time values it all seems okay). Hopefully this (or a better, more elegant fix) will go into 1.2.0: (I removed the old fix, since it does not do any good anymore) diff -r -u -N linux.dist/drivers/char/n_tty.c linux/drivers/char/n_tty.c --- linux.dist/drivers/char/n_tty.c Sun Sep 25 00:41:14 1994 +++ linux/drivers/char/n_tty.c Sun Sep 25 21:44:35 1994 @@ -453,6 +453,8 @@ goto handle_newline; } if (c == EOF_CHAR(tty)) { + if (tty->canon_head != tty->read_head) + set_bit(TTY_PUSH, &tty->flags); c = __DISABLED_CHAR; goto handle_newline; } @@ -718,24 +720,6 @@ *nr -= n; } -/* - * Called to gobble up an immediately following EOF when there is no - * more room in buf (this can happen if the user "pushes" some - * characters using ^D). This prevents the next read() from falsely - * returning EOF. - */ -static inline void gobble_eof(struct tty_struct *tty) -{ - cli(); - if ((tty->read_cnt) && - (tty->read_buf[tty->read_tail] == __DISABLED_CHAR) && - clear_bit(tty->read_tail, &tty->read_flags)) { - tty->read_tail = (tty->read_tail+1) & (N_TTY_BUF_SIZE-1); - tty->read_cnt--; - } - sti(); -} - static int read_chan(struct tty_struct *tty, struct file *file, unsigned char *buf, unsigned int nr) { @@ -744,6 +728,9 @@ unsigned char *b = buf; int minimum, time; int retval = 0; + int size; + +do_it_again: if (!tty->read_buf) { printk("n_tty_read_chan: called with read_buf == NULL?!?\n"); @@ -858,7 +845,6 @@ put_fs_byte(c, b++); if (--nr) continue; - gobble_eof(tty); break; } if (--tty->canon_data < 0) { @@ -896,7 +882,14 @@ current->state = TASK_RUNNING; current->timeout = 0; - return (b - buf) ? b - buf : retval; + size = b - buf; + if (size && nr) + clear_bit(TTY_PUSH, &tty->flags); + if (!size && clear_bit(TTY_PUSH, &tty->flags)) + goto do_it_again; + if (!size && !retval) + clear_bit(TTY_PUSH, &tty->flags); + return (size ? size : retval); } static int write_chan(struct tty_struct * tty, struct file * file, diff -r -u -N linux.dist/include/linux/tty.h linux/include/linux/tty.h --- linux.dist/include/linux/tty.h Wed Aug 10 09:26:44 1994 +++ linux/include/linux/tty.h Wed Aug 10 09:26:44 1994 @@ -247,6 +247,7 @@ #define TTY_EXCLUSIVE 3 #define TTY_DEBUG 4 #define TTY_DO_WRITE_WAKEUP 5 +#define TTY_PUSH 6 #define TTY_WRITE_FLUSH(tty) tty_write_flush((tty)) ------------------------------ From: hstrong@eng1.uconn.edu (Hugh Strong) Subject: Re: ext2fs vs. Berkeley FFS Date: 10 Oct 1994 14:50:20 GMT Michael Bischoff (mbi@math.nat.tu-bs.de) wrote: : In article adc@bach.coe.neu.edu (Albert D. Cahalan) writes: : I think 3 or 4 bits could be spared for extended attributes like this: : 0 flat file (default) : 1 NEW record file : 2 Mac file : 3 NT file : 4 OS/2 file : 5 DOS executable : 6 Windows executable : 7 etc. : Hi, : this would be against the philosophy of UNIX: less is more! : The success and flexibility of UNIX is due to the fact that : all I/O is done through files: ordinary files, hard disks, printers : are all accessable through the same open() and write() calls. : The mess of DOS is that you need special programs if you want : to read A: as entire image. : Michael : -- : EOI : ---------------------------------------------------------------------------- : Michael Bischoff e-mail: mbi@mo.math.nat.tu-bs.de : Abt. Mathematische Optimierung or m.bischoff@tu-bs.de : Inst. Angewandte Mathematik or on.bischoff@zib-berlin.de : Pockelsstrasse 14 Tel. +49-531-391-7555, Fax: +49-531-391-4577 : 38106 Braunschweig Germany I think some people may misunderstand what I an suggesting. Introducing a new set of namespaces like A:,C: and LPT1: would be an intolerable barbarity. Certainly no one accustomed to UNIX would think of using it. But what about extending the semantics of existing calls? This has occured many times in the UNIX world. This is precisely what happens every time someone writes a driver with a new ioctl() call. For instance, to open the main (data) fork of a file, one might write fd = open("MyDataFile",O_RDONLY); The icon (for a window manager) for the file could be accessed by the following call. fd1 = open("MyDataFile:ICON",O_RDONLY); The state of an editing session on the file could be saved in yet another fork fd2 = open("MyDataFile:EDITSTATE",O_RDONLY); It is the FILESYSTEM code that grasps the semantics of what we are doing, not other parts of the kernel. If some of this functionality can be exported to user space, so much the better. I believe that NTFS handles extended attributes in a similar way. I don't think that there's any need for a single filesystem to understand more than one type of multifork FS. Support for other forkful filesystems should be in the code for that fs. A common interface is important, so that fork-aware versions of cp, rm, etc can move data between different filesystems without too much hassle. -- Hugh Strong hstrong@eng1.uconn.edu ------------------------------ From: iialan@iifeak.swan.ac.uk (Alan Cox) Subject: Re: BSD/386 vs. Linux Performance Date: Mon, 10 Oct 1994 10:43:08 GMT In article lalbers@dozy.hrz.uni-bielefeld.de ( Leif Albers) writes: >The main result was (if I'm right): > If you have a small (harddisk and memory), stand-alone machine -- >use Linux. > If you have a machine connected to a network with much network >travel -- use BSD. Both of these are out of date. With its shared libs and clean up BSD runs as well as Linux on small machines and the net code in Linux is now solid. Everyone around here uses Linux mostly because a) Everyone else does b) It's easier to get hold of (in the UK its made a major UK CD magazine [which seems to have immediately sold out everywhere]. c) DOS emulation, and being able use iBCS2 binaries. I think a) probably has the most bearing. Alan -- ..-----------,,----------------------------,,----------------------------,, // Alan Cox // iialan@www.linux.org.uk // GW4PTS@GB7SWN.#45.GBR.EU // ``----------'`----------------------------'`----------------------------'' ------------------------------ ** FOR YOUR REFERENCE ** The service address, to which questions about the list itself and requests to be added to or deleted from it should be directed, is: Internet: Linux-Development-Request@NEWS-DIGESTS.MIT.EDU You can send mail to the entire list (and comp.os.linux.development) via: Internet: Linux-Development@NEWS-DIGESTS.MIT.EDU Linux may be obtained via one of these FTP sites: nic.funet.fi pub/OS/Linux tsx-11.mit.edu pub/linux sunsite.unc.edu pub/Linux End of Linux-Development Digest ****************************** .