Re: Server X process taking 250 Mo - leaking ?
On Fri, 2002-06-14 at 15:31, Ian D. Stewart wrote: Can anybody else shed some light on the difference between SIZE and RSS? Size is how much memory has been allocated through brk(2). RSS is how much is currently paged in. So, a program can (and some do) brk a lot of memory, thus upping their SIZE, but don't actually use it. The RSS is how much actual RAM the program is taking. signature.asc Description: This is a digitally signed message part
Re: Server X process taking 250 Mo - leaking ?
On 2002.06.18 08:04 Anthony DeRobertis wrote: On Fri, 2002-06-14 at 15:31, Ian D. Stewart wrote: Can anybody else shed some light on the difference between SIZE and RSS? Size is how much memory has been allocated through brk(2). RSS is how much is currently paged in. So, a program can (and some do) brk a lot of memory, thus upping their SIZE, but don't actually use it. The RSS is how much actual RAM the program is taking. Thanx Anthony. According to 'man 2 brk', brk sets the end of the data segment to the value speciĀ fied by end_data_segment, when that value is reasonable, the system does have enough memory and the process does not exceed its max data size (see setrlimit(2)). So, if I'm understanding this correctly, SIZE indicates how much memory has been reserved for the application (and therefor not available to other applications), while RSS is the amount of memory currently being used by the application. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Server X process taking 250 Mo - leaking ?
So, if I'm understanding this correctly, SIZE indicates how much memory has been reserved for the application (and therefor not available to other applications), Nope. By default at least, Linux will overcommit memory. So, you can run 40 different programs all with a 1GB size. brk just basically means that if the program accesses below the boundry, ok. If it accesses above the boundry (and there is nothing else there, e.g., through mmap) then a segfault will occur. /me thinks mmap also counts towards SIZE. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Server X process taking 250 Mo - leaking ?
On 2002.06.13 12:34 Derrick 'dman' Hudson wrote: On Thu, Jun 13, 2002 at 11:14:48AM +0200, Jerome Lacoste wrote: | I have the following in top: | | PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND | 532 root 16 -10 289M 32M 7528 S 54.6 6.5 8:35 XFree86 | | I think there is a problem there. XFree86 takes way too much space and | CPU. Have anybody encountered the same problem? The SIZE column is useless. I forget the details why, but it often confuses the uniniated. The RSS is the how much heap the process really has. The %MEM shows how much of your real memory (not swap) the process is using. Can anybody else shed some light on the difference between SIZE and RSS? Thanx, Ian -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Server X process taking 250 Mo - leaking ?
On Thursday 13 June 2002 11:14, Jerome Lacoste wrote: I have the following in top: PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND 532 root 16 -10 289M 32M 7528 S 54.6 6.5 8:35 XFree86 I think there is a problem there. XFree86 takes way too much space and CPU. Have anybody encountered the same problem? [SNIP] I just reinstalled the new NVidia drivers. Any idea? Cheers, Jerome Did you try the XFree86 nv drivers instead of the nvidia drivers? nvidia drivers are known as memory hogs. X on the other hand isn't too small either. -- Embedded Linux -- True multitasking! TWO TOASTS AT THE SAME TIME! -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Server X process taking 250 Mo - leaking ?
On Thu, Jun 13, 2002 at 11:14:48AM +0200, Jerome Lacoste wrote: | I have the following in top: | | PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND | 532 root 16 -10 289M 32M 7528 S 54.6 6.5 8:35 XFree86 | | I think there is a problem there. XFree86 takes way too much space and | CPU. Have anybody encountered the same problem? The SIZE column is useless. I forget the details why, but it often confuses the uniniated. The RSS is the how much heap the process really has. The %MEM shows how much of your real memory (not swap) the process is using. Do you have 256MB or 512MB RAM? (Hint: use the RSS and %MEM to compute approximately how much real memory you have) As for the CPU usage, what are you doing? If I open a large page in galeon (almost full-screen window) and scroll through the whole thing at once, line-by-line, X will peg my CPU. It's one of the effects of doing lots of video, and using vesafb. | Now after a restart it is: | | PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND | 1895 root 9 -10 281M 25M 7528 S0.1 5.0 1:01 XFree86 | | I think it still takes a lot of memory. Is that normal? Looks pretty normal to me (that SIZE column looks odd, but I'd have to research the problems with SIZE to figure out why) : PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND 25655 root 5 -10 43080 26M 14388 S0.1 10.4 72:55 XFree86 (I have 256MB RAM, see how the RSS*%MEM is (approximately) the amount of memory I have?) | I didn't notice that before last week, after I tried to enable true type | fonts. I have true-type fonts enabled, I don't know how much I'm really using them, though. | Appart from xfstt still running I'm not using xfs at all. For a single system, I think it only doubles memory consumption (one copy of font for xfs, one for X). HTH, -D -- Microsoft DNS service terminates abnormally when it receives a response to a dns query that was never made. Fix information: run your DNS service on a different platform. -- bugtraq http://dman.ddts.net/~dman/ pgpTAwxCuBQtN.pgp Description: PGP signature