Issue metadata
Sign in to add a comment
|
Stale .pak files still sometimes happen |
||||||||||||||||||||
Issue descriptionChrome Version: M55 OS: Android 7.0 Package version: 288309102 Package version name: 55.0.2883.91 Device : Moto G4 (athene_f) Build ID: NPJ25.93-14 Build fingerprint: motorola/athene_f/athene_f:7.0/NPJ25.93-14/16:user/release-keys What steps will reproduce the problem? (1) Launch Chrome (2) Navigate to History What is the expected result? History page should load normally with History and other labels What happens instead? Please see attached screenshot and gif. The name of the page is "Certificate Information" The label $1[$2] below works same as "clear browsing data..." Forum : 1. https://productforums.google.com/forum/#!msg/chrome/5Y9T-QfhEPw/wgVoMPg9CAAJ 2. https://productforums.google.com/forum/#!msg/chrome/zCq8UwkLaq0/HLyjJpmfBgAJ Feedback : go/suteu
,
Feb 13 2017
Play Store review about this issue Version code 288309101 Version name 55.0.2883.91 Device Galaxy S6 (zeroflte) OS Android 6.0
,
Feb 17 2017
Another report (53453223322) from another Moto device Package Version Name 55.0.2883.91 Device Model XT1650 URL chrome://history/ com.google.android.webview : VersionCode:288309150;VersionName:55.0.2883.91 Package version: 288309152 Build ID: NCL25.86-11-4-6 With this we have confirmed report from Moto and Samsung devices Attached logs suggest that Chrome is not able to load history on this device (as shown in screenshot) and bonus - this log contain a system scan by Malwarebytes
,
Feb 21 2017
candrada@, can we reproduce this? If not with our basic config, can we see if installing MalwareBytes (I'm guessing this package - https://play.google.com/store/apps/details?id=org.malwarebytes.antimalware&hl=en) helps to reproduce?
,
Feb 22 2017
I am the one who posted the issue 687058 which was merged with this issue,(U can see that above.) To clarify, the issue is in moto G4 plus running android 7.0. device model XT1643. Device's build no. is NPJ25.93-14 I have figured out the origin of this issue. This issue come into existence after the latest system update (till now- 22/02/2017) I also figured out that this issue is seen in devices with chrome version 55.0.2883.91, another devices with chrome version 56.0.2924.87 does not have this issue. Therefore users having this issue can simply update there chrome, it may help.
,
Feb 22 2017
We have spent some time on this issue, but no luck. Also tried installing the MalwareBytes app from playstore and tried working on history. Changed the proxy settings and checked the history but couldn't reproducible. Tomorrow will try different scenarios with Some more devices and update the status here. Devices Tested: Moto G4 / NPJ25
,
Feb 23 2017
Found a few log files where users reported feedback on chrome://history page.
Found a few exceptions in logs like
52429161926
W/System (24135): ClassLoader referenced unknown path: /system/app/Chrome/lib/arm
E/chromium(24135): [ERROR:interface_registry.cc(104)] Failed to locate a binder for interface: autofill::mojom::PasswordManagerDriver
W/System (21830): ClassLoader referenced unknown path: /system/app/Chrome/lib/arm
W/dboxed_process2(21830): type=1400 audit(0.0:2338): avc: denied { getattr } for uid=99002 path="/data/data/com.android.chrome" dev="mmcblk0p48" ino=380 scontext=u:r:isolated_app:s0:c512,c768 tcontext=u:object_r:app_data_file:s0:c512,c768 tclass=dir permissive=0
52471245278
E/KernelUidCpuTimeReader( 2843): Failed to read uid_cputime: /proc/uid_cputime/show_uid_stat (No such file or directory)
E/chromium(15205): [ERROR:interface_registry.cc(104)] Failed to locate a binder for interface: autofill::mojom::PasswordManagerDriver
E/AndroidRuntime( 7622): android.content.ActivityNotFoundException: Unable to find explicit activity class {com.android.chrome/org.chromium.chrome.browser.preferences.website.ManageSpaceActivity}; have you declared this activity in your AndroidManifest.xml?
W/System (16381): ClassLoader referenced unknown path: /system/app/Chrome/lib/arm
52471388606
E/KernelUidCpuTimeReader( 2843): Failed to read uid_cputime: /proc/uid_cputime/show_uid_stat (No such file or directory)
E/chromium(16381): [ERROR:interface_registry.cc(104)] Failed to locate a binder for interface: autofill::mojom::PasswordManagerDriver
W/System (16381): ClassLoader referenced unknown path: /system/app/Chrome/lib/arm
52849390446
E/KernelUidCpuTimeReader( 1459): Failed to read uid_cputime: /proc/uid_cputime/show_uid_stat (No such file or directory)
W/System (27861): ClassLoader referenced unknown path: /system/app/Chrome/lib/arm
W/dboxed_process8(27861): type=1400 audit(0.0:102): avc: denied { getattr } for uid=99015 path="/data/data/com.android.chrome" dev="mmcblk0p48" ino=892 scontext=u:r:isolated_app:s0:c512,c768 tcontext=u:object_r:app_data_file:s0:c512,c768 tclass=dir permissive=0
,
Mar 3 2017
Another report I am facing the same issue and suspecting it to be hacked. Also couple of observations a)When i try to make payment using android app initially payment page displayed blank and did not allow.b)when i copy paste any data,sometimes it comes is different language other than English.c)And finally when i visit payment linked pages in the web, i see Indonesia as country displaying in my network and interestingly calli of the language which i mentioned earlier is matching.
,
Mar 7 2017
Another report I have a similar issue. When I open Chrome History on Android I now see "Emirate" at the top of the screen, with "zip code" at the bottom of the screen. When I drag down the URL bar, I am told to "Update to verify your $1" in the hidden, under browser area. I don't use Apps or do anything other than simple Chrome searches on my Samsung Galaxy Note 4 phone, but I do tether for home Internet access, so this is a real concern. This issue began after my last Verizon software update.
,
Mar 9 2017
Tried Moto G4 / N and also on Samsung Galaxy Note 4 with below scenarios: Autofill payments and payments, tethering from other mobile through data connection and tried as per Comment# 10 but no luck in the reproducibility. We will monitor this issue and update if we get any repro steps from our end.
,
Mar 12 2017
I just updated my LG Aristo 7.0 and have the same issue as jainabhi...$1.
,
Mar 13 2017
Okay, I fixed it. I went to the Google Play Store, updated Chrome Browser and it is back to normal. That was why it showed it was not compatible - because the old Chrome Browser is not compatible with the new update. Update your Chrome Browser and that should solve your problem.
,
Mar 14 2017
Help I have the same message on my LG Aristo
,
Mar 14 2017
jennifer@, can you please visit chrome://version and send us a screenshot of that page? We're trying to identify where this bug exists, and having detailed version data is very useful in doing so.
,
Mar 24 2017
Possible this is related to bug 673458 . That issue was about the sdcard breaking our when-to-extract logic, while this one (at least #6), is about system updates (or maybe backup/restore?). Bottom line, sounds like we need a better way to detect when the pak file is stale. I'll take this on.
,
Mar 24 2017
,
Apr 5 2017
Issue 708369 has been merged into this issue.
,
Apr 12 2017
Issue 710879 has been merged into this issue.
,
Apr 14 2017
agrieve@, since there are bugs being duped in here on a somewhat regular basis, can we try to fix this for M59? Or is that infeasible? Tentatively tagging as such, since I do think this is important, but I'm happy to discuss if you think this is too little time or you have other priorities.
,
Apr 18 2017
I do think this is really important and plan on working on it this week. Likely won't have a fix until next week. I also don't have a good idea yet as to how big a change this will entail, so I can't say as of yet whether it would be a good idea to merge into the now-cut branch.
,
Apr 18 2017
Reminder as to what our current extraction logic looks like: - https://cs.chromium.org/chromium/src/base/android/java/src/org/chromium/base/ResourceExtractor.java - Store versionCode in SharedPreferences. - When versionCode changes, delete all existing .pak files. - Check for the existence of the files to extract on disk - If any of them are missing, extract them. Main issues with above: - Files can be corrupt if app crashes mid-extraction - Odd cases of backup / restore, or moving app to sdcard can cause SharedPreferences to be correct, but .pak to be old Some ideas for how to fix: 1. Build ID as pak file entry - Have each repack() step add an entry containing the build id as resource #0 - Have resource_bundle.cc check that resource #0 is current when opening any .pak file - Add a JNI hook that resource_bundle.cc can call to re-kick extraction when necessary 2. Build ID as pak header entry - Same as #1, except put build id in .pak file header rather than as an entry. This will require rev'ing the pak file version 3. Check the file length of each file when seeing if it's up-to-date - This would work 99.99% of the time, but at this point we're looking for a perfect solution... 4. Encode build-id/versionCode in file names - E.g. en-US.pak would become en-US.pak@$VERSION_CODE when extracted (keep existing name within .apk) - Trigger deletion of old files when any file is found to need extraction. - Remove logic for storing current version in SharedPreferences - Have resource_bundle.cc Although #1 and #2 seem a bit more elegant to me, #4 seems like it will do the trick and is simpler (doesn't have a "retrigger extraction" case, just makes the initial "extract what's needed" step always work.
,
Apr 18 2017
Note: To fix the corrupted-during-extraction case, we should just extract to a temp file then rename.
,
Apr 18 2017
Exactly what will need to change for #4: https://cs.chromium.org/chromium/src/base/android/build_info.h * Move ApkVersion logic from ResourceExtractor.java to BuildInfo.java https://cs.chromium.org/chromium/src/ui/base/resource/resource_bundle_android.cc * LoadFromApkOrFile() will need to add the custom suffix (obtained from build_info.h) https://cs.chromium.org/chromium/src/ui/base/resource/resource_bundle.cc * GetLocaleFilePath() will need to do likewise. The biggest hiccup is that currently tests sideload the .pak files rather than bundling them in the .apk and extracting them. This is also an unknown to me right now effort-wise. I'll dig into what would be required to bundle test .pak files rather than side-load them.
,
Apr 19 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/4c8239ac8c3c3dd581963aeb6e5dcb2e381d0b58 commit 4c8239ac8c3c3dd581963aeb6e5dcb2e381d0b58 Author: agrieve <agrieve@chromium.org> Date: Wed Apr 19 14:19:42 2017 Android: ResourceExtractor.java: Extract files atomically Extracts to a temporary and then moves them into place. The goal is to avoid half-extracted files (e.g. upon crash or power failure). The extractor does not validate them on start-up. BUG= 691719 Review-Url: https://codereview.chromium.org/2828583002 Cr-Commit-Position: refs/heads/master@{#465592} [modify] https://crrev.com/4c8239ac8c3c3dd581963aeb6e5dcb2e381d0b58/base/android/java/src/org/chromium/base/ResourceExtractor.java
,
Apr 19 2017
Issue 712993 has been merged into this issue.
,
Apr 20 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/a9cdba46ce4367c96aa7c9a0655295a5b8de5992 commit a9cdba46ce4367c96aa7c9a0655295a5b8de5992 Author: agrieve <agrieve@chromium.org> Date: Thu Apr 20 18:03:29 2017 Android: Add version suffix to extracted .pak files This will (hopefully) guarantee that we never use a stale .pak file, and actually simplifies the triggering & clean-up logic somewhat (e.g. no longer uses SharedPreferences). BUG= 691719 Review-Url: https://codereview.chromium.org/2828823002 Cr-Commit-Position: refs/heads/master@{#466067} [modify] https://crrev.com/a9cdba46ce4367c96aa7c9a0655295a5b8de5992/base/android/build_info.cc [modify] https://crrev.com/a9cdba46ce4367c96aa7c9a0655295a5b8de5992/base/android/build_info.h [modify] https://crrev.com/a9cdba46ce4367c96aa7c9a0655295a5b8de5992/base/android/java/src/org/chromium/base/BuildInfo.java [modify] https://crrev.com/a9cdba46ce4367c96aa7c9a0655295a5b8de5992/base/android/java/src/org/chromium/base/ResourceExtractor.java [modify] https://crrev.com/a9cdba46ce4367c96aa7c9a0655295a5b8de5992/ui/base/resource/resource_bundle.cc
,
Apr 20 2017
The two attached commits should completely fix the bug. Will leave this open for comment about whether we want to merge to M59.
,
Apr 20 2017
,
Apr 25 2017
,
Apr 25 2017
Your change meets the bar and is auto-approved for M59. Please go ahead and merge the CL to branch 3071 manually. Please contact milestone owner if you have questions. Owners: amineer@(Android), cmasso@(iOS), gkihumba@(ChromeOS), Abdul Syed@(Desktop) For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
,
Apr 27 2017
This bug requires manual review: We don't branch M60 until 2017-05-25. Please contact the milestone owner if you have questions. Owners: amineer@(Android), cmasso@(iOS), josafat@(ChromeOS), bustamante@(Desktop) For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
,
Apr 28 2017
This issue has been approved for a merge. Please merge the fix to any appropriate branches as soon as possible! If all merges have been completed, please remove any remaining Merge-Approved labels from this issue. Thanks for your time! To disable nags, add the Disable-Nags label. For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
,
Apr 28 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/bf64cf69693a7a7611f1f484ad93315bf4b34d2c commit bf64cf69693a7a7611f1f484ad93315bf4b34d2c Author: Andrew Grieve <agrieve@chromium.org> Date: Fri Apr 28 15:27:35 2017 Android: ResourceExtractor.java: Extract files atomically Extracts to a temporary and then moves them into place. The goal is to avoid half-extracted files (e.g. upon crash or power failure). The extractor does not validate them on start-up. BUG= 691719 Review-Url: https://codereview.chromium.org/2828583002 Cr-Commit-Position: refs/heads/master@{#465592} (cherry picked from commit 4c8239ac8c3c3dd581963aeb6e5dcb2e381d0b58) Review-Url: https://codereview.chromium.org/2847923004 . Cr-Commit-Position: refs/branch-heads/3071@{#293} Cr-Branched-From: a106f0abbf69dad349d4aaf4bcc4f5d376dd2377-refs/heads/master@{#464641} [modify] https://crrev.com/bf64cf69693a7a7611f1f484ad93315bf4b34d2c/base/android/java/src/org/chromium/base/ResourceExtractor.java
,
Apr 28 2017
,
Apr 28 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/0009f4a2703ca2b72878f68865714c9c1ea2e677 commit 0009f4a2703ca2b72878f68865714c9c1ea2e677 Author: Andrew Grieve <agrieve@chromium.org> Date: Fri Apr 28 15:31:16 2017 Android: Add version suffix to extracted .pak files This will (hopefully) guarantee that we never use a stale .pak file, and actually simplifies the triggering & clean-up logic somewhat (e.g. no longer uses SharedPreferences). BUG= 691719 Review-Url: https://codereview.chromium.org/2828823002 Cr-Commit-Position: refs/heads/master@{#466067} (cherry picked from commit a9cdba46ce4367c96aa7c9a0655295a5b8de5992) Review-Url: https://codereview.chromium.org/2845163006 . Cr-Commit-Position: refs/branch-heads/3071@{#294} Cr-Branched-From: a106f0abbf69dad349d4aaf4bcc4f5d376dd2377-refs/heads/master@{#464641} [modify] https://crrev.com/0009f4a2703ca2b72878f68865714c9c1ea2e677/base/android/build_info.cc [modify] https://crrev.com/0009f4a2703ca2b72878f68865714c9c1ea2e677/base/android/build_info.h [modify] https://crrev.com/0009f4a2703ca2b72878f68865714c9c1ea2e677/base/android/java/src/org/chromium/base/BuildInfo.java [modify] https://crrev.com/0009f4a2703ca2b72878f68865714c9c1ea2e677/base/android/java/src/org/chromium/base/ResourceExtractor.java [modify] https://crrev.com/0009f4a2703ca2b72878f68865714c9c1ea2e677/ui/base/resource/resource_bundle.cc
,
Apr 28 2017
|
|||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||
Comment 1 by jainabhi...@chromium.org
, Feb 13 2017