Closed Bug 192546 Opened 22 years ago Closed 22 years ago

Page includes grabage including HTTP headers, Image data

Categories

(Core :: Networking: HTTP, defect, P2)

x86
Windows XP
defect

Tracking

()

RESOLVED DUPLICATE of bug 195746
mozilla1.4alpha

People

(Reporter: teilo+bugzilla, Assigned: darin.moz)

References

()

Details

(Whiteboard: [pipelining])

Attachments

(5 files)

User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.3b) Gecko/20030130 Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.3b) Gecko/20030130 Going to the above URL mozilla sometimes displays the server response to requests not the HTML layout. Reproducible: Always Steps to Reproduce: 1. Clear cache 2. Restart Mozilla 3. Go to http://www.mirada-solutions.com/ 4. in the page go to "Contact Us" -> "How to Reach Us" Actual Results: Page includes grabage including HTTP headers, Image data etc.. Expected Results: Page renders
Screenshot of browser window. (text wouldn't copy/paste)
Sorry but it works for me, current CVS based trunk build, WinXP-Sp1.
possible dupe of 140107 but this page worked with 1.3a with pipelining on. Keywords -> Regression?
Reproducable 100% here, using mozilla-win32-svg-GDI-mathml.zip as of today
Sorry, again, but it works for me. With / without http pipelining enabled.
do you use any proxy ? http://webtools.mozilla.org/web-sniffer/ shows the Web server doesn't return any unusual HTTP headers.
Olivier: I don't use any proxy.
To reproduce follow the steps at the top. Going *directly* to the page is okay for me.
This is worksforme with win2k build 20030209.. Reporter: Can you please uphrade to the latest build and clear the disk cache ? And you don't use a firewall like Zone Alarm with ad-blocking ?
Matti, Confused, this is the latest, see comment #4 About shows,Gecko/20030130 but Window Title bar shows 2002020800 *BUT* downloaded mozilla-win32-svg-GDI-mathml.zip today (10 February 2003, 10:58:48) from http://ftp.mozilla.org/pub/mozilla/nightly/latest-trunk/ Also, Step one (comment 0) would invlove clearing the cache ;-) There is a firewall (XP personal firewall) but this was running with 1.3a and 1.3a does not have this bug, also the firewall does not provide any HTTP filtering. WinXP Pro SP1 + misc windows update hotfixes Non SVG-MATHML build is on its way down a 40kbps modem...
Still reproducable with Gecko/20030210. 1)Clear Cache 2)Go to http://www.mirada-solutions.com/ 3)in the page go to "Contact Us" -> "How to Reach Us" (can click before page is 100% loaded) 4)goto 2 while page displays correctly. Note I'm on a 56k modem and the page takes a while to load View Page source shows correct source Hitting Reload while on the corrupted page results in a blank page view source shows only <html><body></body></html> Cache: Compare every time I visit the page Cache Folder: Profile Folder stored on local disc Link Prefetching: on Networking HTTP/1.1 Pipelining enabled Keepalive enabled
Changed settings to disable keepalive & pipelining and selectevly re-enabled. Now the settings are back where I started and can't reproduce it anymore?!
Status: NEW → RESOLVED
Closed: 22 years ago
Resolution: --- → WORKSFORME
VERIFIED/WFM: Win 98, Mozilla 1.3b My experience is that clearing the cache did not work in all cases. Doing a new profile test would have been a better troubleshooting step here. Possibly a bad cache entry was properly cleared as you kept making changes.
Status: RESOLVED → VERIFIED
Summary: moz renders Server response → Page includes grabage including HTTP headers, Image data
Reopenning, as created a new profile and saw this *sometimes* after I enabled pipelining (never before). The timing seems to be critical, following the link when the page is just rendered before the images are rendered seems to do trigger the problem. Server claims to be Apache 1.3.27 (latest stable 1.3) so I wouldn't have thought that it contains pipelining bugs, but someone with more knowledge of this may wish to mark this as a duplicate. I would be more inclined to think this was a subtle threading issue problem (but im not the developer!). I have not seen a similar problem with other sites (yet ;-) )
Status: VERIFIED → REOPENED
Resolution: WORKSFORME → ---
WORKSFORME linux trunk 2003-02-11/08. Sounds like this needs to be tested over a modem link. James: can you capture a HTTP log for us? all you need to do is set some environment variables before running mozilla. here's the steps: 1- make sure mozilla is not running (exit from quick launch if necessary) 2- open a DOS command prompt and type the following commands: C:\> set NSPR_LOG_MODULES=nsHttp:5 C:\> set NSPR_LOG_FILE=C:\http.txt C:\> cd \path\to\mozilla C:\path\to\mozilla\> .\mozilla.exe 3- clear cache 4- enable pipelining 5- load page 6- exit mozilla completely 7- upload C:\http.txt to this bug report this will help me understand where things are going wrong. THANKS!!
Attached file Log file
Log produced according to above comment. Things to note, took ~15 attempts (clear cache/restart/follow links/) to show up. Initial request to /contact.html resulted in a completly blank page Hitting the Reload button caused the garbage page to appear.
Similar issue on engrish.com. User-Agent: Mozilla/5.0 (Windows; U; Win98; en-US; rv:1.3b) Gecko/20030218 Netcraft says the server is: Apache/1.3.22 (Unix) (Red-Hat/Linux) mod_jk/1.2.0 mod_perl/1.24_01 PHP/4.1.1 FrontPage/5.0.2 mod_ssl/2.8.5 OpenSSL/0.9.6b Steps to reproduce: 1. Clear cache 2. Enable pipelining 3. Go to http://www.engrish.com/ 4. Click on a link on the left (like Snacks) 5. While (or before) the thumbnails are loading, click on one of the thumbnails Actual results: One of the following: - Empty page; Page Info says: URL is OK, type is text/html, size is 0 bytes - Page loads normally, but one of the images is replaced by one of the thumbnails from the previous page. Properties for the replaced image: URL of the image that should be there, size of the image that is displayed. - Page loads, but a bunch of **** is prepended to it, likely image data (including the HTTP header for the html file; in fact on one occasion I saw two HTTP headers, one of them was clearly a jpeg). Sometimes one of the images is replaced by one of the thumbnails from the previous page. Page info reports a much larger size than what it would normally report for that page. - One of the thumbnails from the previous page is loaded; Page Info says: URL is of the html that should have been displayed, the rest of the reported data is of the picture. Sometimes, when one of the thumbnails from the previous page is displayed on the resulting page, when going back to the thumbnails page, the "broken image" placeholder is displayed instead of that thumbnail. Sometimes, when I've played long enough with this, when I go back to the home page (using the back button) most of the links are broken, e.g. right now clicking on "Chocolate" (which leads to the url http://www.engrish.com/chocolate/ ) I get one of the thumbnails from the gallery page I've come "back" from. Properties says Location is http://www.engrish.com/chocolate/ , the rest of the information seems to be that of the image. Clicking on some of the other links results in a download window with the right URL but content-type: application/octet-stream. None of this happens when I disable pipelining. Expected results: Page should load normally.
Mozilla/5.0 (Windows; U; Win98; en-US; rv:1.3b) Gecko/20030219 1. enabled pipelining ( keep-alive was enabled ) 2. www.engrish.com 3. selected chocolate 4. got garbage by selecting some of the thumbnails as described in comment above. Then I clicked shift-reload and waited for the page to reload. Didn´t succeed to click on the first images, but clicking on lower images often produced garbage. Then I only clicked Back / Reload and clicked another image, also often got garbage. Attachment is made this way, original link: http://www.engrish.com/chocolate/source/menspocky.html Often got headers like this on www.siliconstrategies.com
my favourite website producing this kind of bugs. I´m loading the main page, and then opening about four to more then ten articles loading in the background. Often some of the links didn´t load, or showed some javascript source ( Dr. clue´s ), sometimes something is distorted, if pipelining is enabled. This screenshot is like the browser and the article from feb 3rd, and illustrates effects at the end of the loaded file. But distortions always are at the beginning of the loaded file, as seen above in my engrish screenshot.
I'm seeing this too sometimes with pipelining enabled, mostly as described in comment 17. For some news to this, I recorded a log of a failed 'snack' attack. ;) What I did: -Enable HTTP logging -Cleared cache -Go to http://www.engrish.com/ -Before load finished, clicked on link http://www.engrish.com/snacks/ -Mozilla offered me to save some application/octet-stream, which I did. I've uploaded the HTTP log and the saved file with its suggested save name. The file contains three HTTP responses for gif images which I verified to be the following gif's from the front page (in this order): http://www.engrish.com/images/categories.gif http://www.engrish.com/images/bottoms.gif http://www.engrish.com/images/submissions.gif ... and finally the originally requested html page together with its HTTP header. I think its some bug related to aborting loading a page, and then reusing this HTTP connect for further requests.
Status: REOPENED → ASSIGNED
Priority: -- → P2
Whiteboard: [pipelining]
Target Milestone: --- → mozilla1.4alpha
i'm able to repro some bad behavior on the site as well. thanks for the log files folks. i'll try to nail this one during the 1.4 cycle.
the site seems to be working now.. this was probably a duplicate of bug 195746. *** This bug has been marked as a duplicate of 195746 ***
Status: ASSIGNED → RESOLVED
Closed: 22 years ago22 years ago
Resolution: --- → DUPLICATE
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: