Open Hub
Open Hub

 Settings |  Report Duplicate

0
I Use This!
Activity Not Available

Commits : Listings

Analyzed 4 months ago. based on code collected 4 months ago.
Jul 20, 2025 — Jul 20, 2026
Commit Message Contributor Files Modified Lines Added Lines Removed Code Location Date
Updated test_virgo_malloc.c syscall client for virgo_get() with a char[] instead of char* for received value.
ka_shrinivaasan
More... almost 13 years ago
kern.log for working virgo_malloc(),virgo_set() and virgo_get() system calls: After commenting sock_release all the 3 calls work and the VIRGO unique id to address mapping and inversion also work end to end from syscall invocation in test_virgo_malloc.c client to VIRGO mempool driver.
ka_shrinivaasan
More... almost 13 years ago
All kernel sock_release() invocations have been commented due to kernel panics. Should be uncommented later.
ka_shrinivaasan
More... almost 13 years ago
Uncommented virgo_get syscall path tests
ka_shrinivaasan
More... almost 13 years ago
kern.log with working virgo_malloc and virgo_set syscalls path (timestamp 11:36)
ka_shrinivaasan
More... almost 13 years ago
VIRGO mempool command creations have been modified as: - allocate a temp buffer and concatenate the virgo mempool command messages with parameters from system call invocation with strcpy and strcat without kstrdup() - copy the temp buffer to the buf that is sent as payload to VIRGO mempool driver. - Earlier code was strcat()-ing to a const char* which could have crashed. With the above no const char* are created. - After this code change virgo_malloc and virgo_set calls are working , but sock_release crashes as usual in virgo_set which will be commented in next commits. - But still there is some unpredictability in connection behaviour and sometimes -32, -107 errors among others do occur. But today's kern.log (which is added to repository) with timestamp in the range 11:36 has above working virgo_malloc and virgo_set. - The Virgo Unique ID generation from VIRGO kernel address in virgo_malloc and inverting it to virgo_address within virgo_set system calls are working as in logs with memcpy() bypassing copy_from_user() which does lot of sanity checks which could have crashed in last months' syscall path testing
ka_shrinivaasan
More... almost 13 years ago
Today's test kern.log for system call path for virgo_malloc and virgo_set system calls: virgo_malloc syscall works and virgo_set has a crash in strcat()
ka_shrinivaasan
More... almost 13 years ago
Ongoing bugfix for syscall path issues like -32 and -107 error codes, and sometimes -512 error codes: - modified all char* buf to char buf[BUF_SIZE] - explicit strcpy() to buf from virgo mempool commands. - lots of guard clauses to prevent negative chunk sizes in virgo_malloc - commented str_to_addr() invocation - more debug printk(s) - all kernel socket datastructures have been moved into the while loop of the virgo_malloc
ka_shrinivaasan
More... almost 13 years ago
Updated VIRGO Design Document
ka_shrinivaasan
More... almost 13 years ago
Laptop Screenshots for telnet path test for working virgo_cloud_malloc, virgo_cloud_set and virgo_cloud_get commands that allocate, set data and retrieve it from a remote cloud node, are added to repository. The buffer is printed that is returned from VIRGO mempool driver.This is almost similar to memcached and is a minimum key-value store functionality. This does not do any VIRGO unique id obfuscation that is only implemented in system call path.Though the telnet path is working as in screenshots, the system call path is plagued by multiple kernel socket related failures that are random and erratic that do not look like system call code level errors.
ka_shrinivaasan
More... almost 13 years ago
Updated buildscript.sh
ka_shrinivaasan
More... almost 13 years ago
Updated intptrtest binary
ka_shrinivaasan
More... almost 13 years ago
updated test_virgo_malloc binary
ka_shrinivaasan
More... almost 13 years ago
Commented debug prints and virgo_get syscall.
ka_shrinivaasan
More... almost 13 years ago
changed char* to const char*.
ka_shrinivaasan
More... almost 13 years ago
kern.log with debug printk()s added to print the size and other variables in functions. The size value is printed negative which points to a serious kernel heap corruption.
ka_shrinivaasan
More... almost 13 years ago
Changed all GFP_ATOMIC to GFP_KERNEL allocations in all kmalloc()s. Added more debug printk(s) to analyze the mysterious behaviour in size and sum_alloc_size. As in kern.log committed along with this, the size parameter becomes negative without any reason while the mempool driver correctly allocates and sends the address to virgo_malloc syscall client. This looks like a serious corruption within kernel heap.
ka_shrinivaasan
More... almost 13 years ago
Important kern.log during testing of virgo mempool syscall path that has quite a few confusing outputs for same kernel and driver builds: 1.Sometimes the broken pipe error occurs (-107 and -32) and sometimes doesn't during kernel socket connect from syscall. 2.Kernel panic occurs sometimes and sometimes doesn't. 3.Most erratic is the random execution of if(sum_alloc_size==size) which is not executed at all sometimes though size is always 1000 and this causes the loop to continue while it should have broken. which is difficult to explain without suspecting lowlevel microcode gimmicks. 4.The simple_strtoll works well in normal mode and erratic in recovery mode. 5.In recovery mode telnet path works well while it is erratic in normal mode. 6.In summary the complete test logs for November 24 to November 26 is quite confusing and difficult to explain. 7.Output behaviours change after multiple executions for same code.
ka_shrinivaasan
More... almost 13 years ago
Updated VIRGO Design Document
ka_shrinivaasan
More... almost 13 years ago
telnet path tests for virgo_cloudexec_mempool driver for virgo_malloc, virgo_get and virgo_set commands. Dt.11/11/2013 had kernel panics which are not in kern.log of 22November2013. This is probably due to commenting the kernel socket shutdown similar to virgo_malloc syscall.
ka_shrinivaasan
More... almost 13 years ago
Updated CV in ProductOwnerProfile
ka_shrinivaasan
More... almost 13 years ago
Updated VIRGO Design Document
ka_shrinivaasan
More... almost 13 years ago
Updated VIRGO Design Document
ka_shrinivaasan
More... almost 13 years ago
Updated build generated source files and kernel object files.
ka_shrinivaasan
More... almost 13 years ago
Commented kernel socket shutdown to make it similar to virgo_malloc.c syscall client to investigate kmem_cache_trace_alloc panics after each successful execution of virgo_malloc, virgo_set and virgo_get commands in telnet path.
ka_shrinivaasan
More... almost 13 years ago
kern.log for working virgo_malloc, virgo_set and virgo_get calls via telnet path. But there are lot of unrelated kmem_cache_trace_alloc kernel panics after each successful execution of virgo_malloc,virgo_set and virgo_get command.
ka_shrinivaasan
More... almost 13 years ago
Updated CV in ProductOwnerProfile
ka_shrinivaasan
More... almost 13 years ago
typo fixed.
ka_shrinivaasan
More... about 13 years ago
Updated VIRGO Design Doc
ka_shrinivaasan
More... about 13 years ago
Updated VIRGO Design Document
ka_shrinivaasan
More... about 13 years ago