Bug 1925204
| Summary: | glibc: printf of long double with a value of 0 results in "nan" [rhel-7.9.z] | ||
|---|---|---|---|
| Product: | Red Hat Enterprise Linux 7 | Reporter: | Oran Agra <oran> |
| Component: | glibc | Assignee: | glibc team <glibc-bugzilla> |
| Status: | CLOSED ERRATA | QA Contact: | qe-baseos-tools-bugs |
| Severity: | urgent | Docs Contact: | |
| Priority: | urgent | ||
| Version: | 7.9 | CC: | ashankar, ayadav, biholmes, brandfbb, brclark, codonell, dj, elgueromexicano, eric.wood, fweimer, jrd-rhbz, jreznik, juqiao, kahara, knweiss, mails.bugzilla.redhat.com, mnewsome, pasik, pasteur, paul.f.roberts.jr, pfrankli, rhel, sipoyare, skolosov, sreber, ziegleka |
| Target Milestone: | rc | Keywords: | Triaged, ZStream |
| Target Release: | --- | ||
| Hardware: | x86_64 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | glibc-2.17-323.el7_9 | Doc Type: | Bug Fix |
| Doc Text: |
Cause:
The fix for CVE-2020-29573 resulted in a regression in the printf family of functions on i686 and x86_64 when printing zero in long double format.
Consequence:
The printf function would print "nan" for zero on i686 and x86_64.
Fix:
The logic to identify NaNs was updated to ensure that printf family of functions identify NaNs correctly.
Result:
The printf family of functions now print long double zero correctly.
|
Story Points: | --- |
| Clone Of: | Environment: | ||
| Last Closed: | 2021-02-08 15:37:09 UTC | Type: | Bug |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
This bug seems to trigger some serious issues. In my case the glibc-update resulted in a lot of failures in an asterisk-installation where a lot of conditionchecks in the dialplan suddenly returned "nan" instead of "0". I had to downgrade via "yum downgrade glibc glibc-common glibc-devel glibc-headers" to 0:2.17-317.el7. *** Bug 1925306 has been marked as a duplicate of this bug. *** (In reply to Oran Agra from comment #0) > This started failing Redis's CI yesterday (only on CentOS): > https://github.com/redis/redis/runs/1827018214?check_suite_focus=true > so i suppose it's some glibc update that was rolled out by someone. Thank you very much for this report. This is related to the fix for CVE-2020-29573, where we made a change with limited scope to correct the security defect. We are working to correct this issue as quickly as we can. Thank you for your patience. I've edited the description to mention the correct expected results, given that %f does not remove trailing zeros after the decimal point: Expected results: 0 -0 +0.000000 Verified, the reproducer returns 0 and 0.000000. Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (glibc bug fix update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2021:0439 |
Description of problem: After a certain upgrade printf of long double with a value of 0, produces "nan" (printing non-long doubles, or long double with value other than 0 seem fine) Version-Release number of selected component (if applicable): How reproducible: easily reproducible. Steps to Reproduce: $ cat ld2str.c #include <stdio.h> void main() { long double d = 0; printf("%Lg\r\n",d); printf("%Lf\r\n",d); } $ gcc ld2str.c $ ./a.out Actual results: nan nan Expected results: 0 0.000000 Additional info: printf("glibc version: %s\n", gnu_get_libc_version()); glibc version: 2.17 $ rpm -q glibc glibc-2.17-322.el7_9.x86_64 This started failing Redis's CI yesterday (only on CentOS): https://github.com/redis/redis/runs/1827018214?check_suite_focus=true so i suppose it's some glibc update that was rolled out by someone.